Algorithms doi: 10.3390/a19070568

Authors: Hassan Al-Derham Mahitha Veeramachaneni Lu Gao Yunpeng Zhang Jingran Sun Ahmed Senouci Kevin Fu

Post-disaster debris detection is important for rapid damage assessment, emergency response, and recovery planning. However, debris objects in aerial imagery are often fragmented, irregularly shaped, partially occluded, and visually confused with shadows, vegetation, roofs, vehicles, and damaged structures. This study proposes YOLO-SDD, a YOLO-based Shape-Guided Debris Detector built on YOLOv8 for localized debris identification in high-resolution post-disaster aerial imagery. YOLO-SDD combines a high-resolution P2 detection pathway with a shape-guided feature refinement module that uses box-supervised pseudo-mask and pseudo-boundary cues to refine P2-level features before final debris detection. A multi-event aerial imagery dataset was constructed from NOAA Emergency Response Imagery using images collected after hurricanes and a tornado in the United States. The model was evaluated using an image-level split, an event-level holdout test, component-level ablation studies, COCO-style scale-specific evaluation, and multi-seed stability analysis. On the image-level test set, YOLO-SDD achieved a precision of 0.959, recall of 0.933, mAP@50 of 0.970, and mAP@50:95 of 0.755, remaining competitive with larger YOLO-family models at lower computational complexity. In the event-level holdout test, YOLO-SDD achieved an AP@50 of 0.80 and an F1 score of 0.79, outperforming the YOLOv8s baseline and the selected large YOLO-family comparison model. The scale-specific evaluation showed improved AP@50 and recall for small and medium debris groups, while failure cases remained associated with shadows, vegetation, low contrast, and highly fragmented debris. The results indicate that shape-guided P2 refinement can improve localized debris screening under the tested conditions, although broader datasets, workflow integration, and human-in-the-loop validation are still needed before operational deployment.

Algorithms doi: 10.3390/a19070567

Authors: Debtoru Chatterjee Girish Tiwari Niladri Chatterjee

The Colonel Blotto game, formulated by Émile Borel, involves players allocating limited resources to multiple “battlefields” simultaneously, with the winner being the one who allocates more resources to each battlefield. Computation of the Nash equilibrium, including of two-person, zero-sum, mixed strategy Colonel Blotto games have encountered issues of scalability and complexity owing to their PPAD completeness. This paper proposes an algorithm that computes the same value as the Nash equilibrium but cannot be characterized by the Fixed-Point Theorems of Tarski, Kakutani and Brouwer. The reduced complexity of the proposed algorithm is based on dispensing with the need for computing both players’ Nash strategies in Colonel Blotto games. The same algorithm can, therefore, be extended to all two-person, zero-sum games to compute the value of the Nash equilibrium. The theoretical superiority of the proposed algorithm over both LP solvers and another method that computes the same value of the game as its Nash equilibrium by a random assignment of probabilities to the active strategy set of the defending player, is also proposed.

Algorithms doi: 10.3390/a19070566

Authors: Omar Bahou Karim El Moutaouakil Savin Treanţă

This study introduces the ’Black-Backed Jackal Optimization’ (BBJO), a nature-inspired meta-heuristic algorithm designed for complex, non-linear, and high-dimensional search spaces. The fundamental mathematical model of BBJO relies on the opportunistic hunting behavior and survivability strategies of the black-backed jackal (Lupulella mesomelas). We use non-linear energy decrease and adaptive Lévy flight to maintain the equilibrium of the search. This allows the algorithm to scan large areas first, then zoom in with a high degree of precision once it has identified a suitable location. This configuration prevents the algorithm from getting stuck on a suboptimal local solution, which is a frequent danger during searches in complex spaces. BBJO has been validated against 23 standard benchmark functions, demonstrating significantly greater accuracy than Particle Swarm Optimization (PSO) on complex and large-scale search spaces. On fixed-size domains (F21–F23), the BBJO algorithm achieved a 100% success rate with zero standard deviation, surpassing the Grey Wolf Optimizer (GWO) and Differential Evolution (DE), which frequently suffered from structural stagnation. Visual convergence study shows that BBJO efficiently identifies optimal search regions early in the iteration budget, saving time compared to traditional linear decay models. BBJO optimizes fuzzy inference systems (FISs) for two practical applications: autonomous car speed control and industrial furnace regulation. Experimental results indicate that BBJO significantly decreased cumulative penalties and improved steady-state error reduction compared to baseline configurations and established meta-heuristic methods. The results show that BBJO is a reliable and useful technique for engineering optimization.

Algorithms doi: 10.3390/a19070565

Authors: Abhishek Jain Mohammad Atif Siddiqui Tirumalasetty Chiranjeevi Łukasz Knypiński

This paper presents a unified comparative study of various controllers, including proportional–integral–derivative (PID), Fractional-Order PID (FOPID), Internal Model Control (IMC) controllers, and Tilt–Integral–Derivative (TID) controllers, for the regulation of mean arterial blood pressure (MABP). The controllers are optimally tuned by using a single metaheuristic approach, namely the Pelican Optimization Algorithm (POA), ensuring a fair and consistent comparison. The POA optimizes the objective function using standard error indices (ITAE, IAE, and ISE) along with transient characteristics. The aforementioned controllers are then evaluated under varying patient conditions for different patient categories, including sensitive, nominal, and insensitive, and their performance is systematically compared with one another and with the reported methods from the existing literature. The simulation results demonstrate that IMC offers fast settling with minimal overshoot, FOPID improves robustness through fractional dynamics, and the TID controller provides the smoothest transient response and disturbance rejection across all patient categories. The results confirm the effectiveness of advanced control strategies over conventional PID and highlight the potential of POA-tuned TID control for reliable and patient-specific MABP regulation in critical care applications.

Algorithms doi: 10.3390/a19070564

Authors: Haifeng Fu Yaxin Xie Xinlei Xiao Yonghua Han Le Dai

To address the challenges in maize leaf disease detection, including large variation in lesion scales, weak texture of small targets, strong background interference, limited recall ability for blurred lesions, and computational redundancy of conventional detection heads, this paper proposes a lightweight detection algorithm based on an improved YOLOv11n. First, a multi-scale global context kernel attention module is designed, which employs GCKA-bottleneck with large-kernel attention and residual connections to enhance the deep semantic representation of multi-scale lesions. Second, a GSConv-enhanced coordinate multi-receptive attention module is constructed, which combines coordinate position awareness and multi-scale depthwise convolution. Finally, a regression-enhanced depthwise-separable decoupled detection head is proposed to decouple classification and regression tasks, and introduces depthwise separable convolution and a distributed bounding box regression. On a public dataset containing four classes, the improved model achieves an mAP@0.5 of 85.36%, a recall of 83.32%, and a precision of 86.91%, which are 3.46, 27 2.94, and 2.38 percentage points higher than those of the original YOLOv11n, respectively. Meanwhile, GFLOPs and parameter count are reduced by 27.0% and 12.4%, respectively. The proposed algorithm strikes a favorable balance between accuracy, real-time performance, and lightweight design, providing a feasible technical support for field deployment in intelligent agricultural disease monitoring systems.

Algorithms doi: 10.3390/a19070563

Authors: Carlos A. Martínez-Miwa Rocío M. Sánchez-Albores Yojana J. P. Carreón Jorge González-Gutiérrez Mario Castelán

Dried droplet imaging has become an established approach for analyzing complex fluid systems as evaporation patterns preserve physicochemical and structural information. However, spatial variability of these patterns limits the effectiveness of conventional texture descriptors. This work presents a region-aware deep learning algorithm for the classification of dried droplets through spatial decomposition and convolutional learning. Dried droplet images were geometrically standardized prior to patch extraction. The Hough transform supported consistent crown–core localization and patch extraction across the dataset. Each image was divided into twenty localized regions. Twelve represented peripheral patches sampled at 30° intervals. The remaining eight patches were extracted from four central regions using two complementary angular orientations (30° and 90°) per region. This process allowed dataset augmentation and evaluation of regional contributions. Patches were processed using the VGG16 convolutional neural network, following repeated patch-level and droplet-level partitioning strategies designed to prevent data leakage. The proposed algorithm was evaluated on Methotrexate (MTX) droplets prepared with 40%, 60%, and 80% water dilution levels relative to a reference solution. Results show that the proposed strategy increases classification accuracy, outperforming traditional descriptors. These findings demonstrate that spatial decomposition combined with convolutional learning constitutes an effective approach for heterogeneous evaporative pattern classification.

Algorithms doi: 10.3390/a19070562

Authors: Mohammad Abdel-Majeed Iyad Jafar Omar AL-Karadsheh Shorouq Al-Awawdeh Siraj Zabadi Mahdi Flefl

Accurate assessment of radiographic bone loss (RBL) is essential for periodontal diagnosis and staging; however, manual measurement from dental radiographs is labor-intensive, time-consuming and subject to inter- and intra-examiner variability. Existing AI-based methods primarily formulate bone loss assessment as classification, landmark prediction, or direct segmentation of thin anatomical structures, limiting measurement interpretability and robustness. This study proposes clinically interpretable two-phase framework for automated and clinically interpretable RBL estimation from periapical radiographs. The framework explicitly separates anatomical structure recognition from geometric measurement, improving transparency and reducing error propagation. In the first phase, deep learning models segment key anatomical structures, including the crown, root, third root and alveolar bone. In the second phase, a deterministic geometric algorithm extracts clinically relevant landmarks, including the cemento–enamel junction (CEJ), bone crest, and root apex, and computes root length, CEJ–bone crest distance, and radiographic bone loss following established periodontal measurement principles. The framework was evaluated on a curated dataset of annotated radiographs. DS-TransUNet achieved the best segmentation performance. Quantitative evaluation yielded mean absolute errors of 0.81 mm for CEJ–bone crest distance, 0.71 mm for root length, and 5.89% for RBL estimation. Bland–Altman analysis demonstrated minimal systematic bias (−1.03%) and good agreement with expert measurements across different disease severities, supporting the framework’s potential as an objective and clinically applicable tool for periodontal bone loss assessment.

Algorithms doi: 10.3390/a19070561

Authors: Qingye Wang Jiacai Pan Yifeng Zhao Zhihui Hu Zheping Shao Sainan Wang

To address the problems of slow convergence, long planned paths, and excessive turning points in unmanned surface vehicle (USV) path planning under complex environments, this paper proposes a path planning method based on an Enhanced Gold Rush Optimizer (EGRO). A nonlinear adaptive parameter adjustment strategy and a stage-wise dynamic probability mechanism are designed to improve the balance between global exploration and local exploitation at different stages of iteration. In addition, a Gaussian diffusion mechanism combined with a local search operator is introduced to enhance the algorithm’s ability to escape from local optima and reduce the number of path turning points. In the remote-sensing-image-based sea-ice simulation scenario, compared with the conventional GRO, PSO, and GWO algorithms, the maximum observed improvements of EGRO in best fitness, convergence iterations, and the number of path turning points are approximately 31.65%, 51.24%, and 35.00%, respectively. The simulation results indicate that EGRO can provide a feasible swarm-intelligence-based optimization framework for USV path planning. The proposed algorithm can generate feasible paths with relatively shorter lengths and fewer turning points. These characteristics may provide a favorable geometric reference for subsequent trajectory generation and navigation control, thereby highlighting the potential value of EGRO in engineering applications of USV path planning.

Algorithms doi: 10.3390/a19070560

Authors: Byron Wladimir Oviedo-Bayas Stefany Michelle Perachimba Panezo Jorge Humberto Guanin-Fajardo Stalin Daniel Carreño Sandoya

The classification of malicious network-traffic is critical to cybersecurity. However, to the best of our knowledge, no previous studies have performed a comparative analysis of supervised algorithms for classifying malicious traffic, specifically within the network environment of UTEQ, an academic setting with distinctive traffic patterns and security policies. This study compared the performance of four supervised machine learning algorithms (K-Nearest Neighbors, Decision Tree, SVM-RBF, and SVM-Polynomial) using the CRISP-DM methodology. The dataset consisted of 1182 records with 30 variables from Hillstone Networks firewall logs at UTEQ, representing three categories: Normal (74.3%), Botnet_Activity (16.4%), and Other_Malware (9.3%). Preprocessing techniques included SMOTE balancing and Relief-based feature selection (reducing the variables to eight). The area under the curve (AUC) was used as a primary discriminant metric under two complementary one-vs-rest aggregation schemes. Using a support-weighted AUC, K-Nearest Neighbors (k = 7) obtained the highest value (AUC = 0.6147), followed by SVM-Polynomial (0.5846), Decision Tree (0.5724), and SVM-RBF (0.5784), with SVM-RBF obtaining the highest accuracy on the unified eight-feature test set (73.8%). Using a macro-averaged AUC, SVM-Polynomial obtained the highest value (0.6166), closely followed by KNN (0.6133). All AUC values fell within a narrow range (0.57–0.62). A class-wise analysis showed that no single model provides strong discrimination for all traffic classes, and that algorithm selection for operational deployment should be guided by the specific class of interest rather than by a single aggregate metric.

Algorithms doi: 10.3390/a19070559

Authors: Zhaoyang Wang Da Xu Yuze Ma

Complex three-dimensional path planning requires a planner to generate collision-free, short, and smooth paths within limited computation time, but traditional RRT-based methods often suffer from unguided sampling, repeated expansion failures in dense obstacle regions, redundant initial paths, and collision-prone post-processing. To address this problem, this study defines the planning task as efficient path generation in a bounded three-dimensional obstacle space and proposes an environment feedback hybrid sampling bidirectional RRT method integrated with a path-aware improved whale optimization algorithm. In the initial search stage, the algorithm uses the collision rate of each random tree to switch among open-space exploration, heuristic convergence, and blocked region escape sampling. Local obstacle density estimation is further introduced to fuse the sampling direction, goal direction, opposite tree attraction, and obstacle repulsion, while adaptive dual step sizes, backtracking safe step size adjustment, and local rewiring reduce invalid expansions and improve the quality of the first feasible path. In the post-processing stage, the whale optimization algorithm is used to optimize key path nodes rather than all nodes, with path corridor constraints, dynamic fitness weighting, collision repair, elastic band refinement, and B-spline smoothing to shorten the path and improve smoothness while maintaining feasibility. Tested independently 100 times in each of four MATLAB three-dimensional obstacle environments and compared with the best-performing comparison algorithm in each environment, the proposed method reduced planning time by 64.4%, 83.4%, 80.1%, and 39.5%, respectively, and shortened path length by 4.9%, 7.1%, 13.4%, and 10.1%. The success rate reached 100% in the first three environments and 97% in the most complex dense obstacle environment. These results show that the proposed framework improves search efficiency, path quality, and robustness for three-dimensional collision-free path planning under complex obstacle constraints.

Algorithms doi: 10.3390/a19070558

Authors: Efraín Evaristo Díaz Macías José Steven Cordero Bazurto Byron Wladimir Oviedo Bayas

Exact probabilistic inference in Bayesian Networks (BNs) becomes increasingly expensive as network size and structural complexity grow, limiting its applicability in time-sensitive decision-support systems. This study presents a hybrid inference framework that accelerates the deterministic component of Bayesian reasoning by integrating Boolean Satisfiability (SAT) techniques with Bayesian Networks. The proposed approach transforms deterministic conditional probability table (CPT) entries into conjunctive normal form (CNF), enabling SAT-based logical inference over deterministic constraints while preserving the original Bayesian model for probabilistic reasoning. The framework was evaluated on 25 benchmark Bayesian networks using five independent executions per dataset under identical experimental conditions. Performance was assessed through execution time, instrumented operation counts, and inference coverage, with results reported as mean values, standard deviations, and 95% confidence intervals. Experimental results demonstrate substantial reductions in deterministic inference time while maintaining high coverage of deterministic variable assignments across the evaluated benchmarks. Throughout this paper, the reported performance gains refer exclusively to empirical reductions in execution time and instrumented operation counts. They should not be interpreted as evidence of a reduction in the asymptotic computational complexity of exact Bayesian inference, which remains #P-complete in the general case. Rather, the proposed framework provides an efficient mechanism for accelerating deterministic logical inference within Bayesian Networks under the evaluated benchmark conditions.

Algorithms doi: 10.3390/a19070557

Authors: Wonil Choi Klaus Caka Adnan Shah Talha Ali Khan Iftikhar Ahmed Raja Hashim Ali

Reliable traffic sign classification is essential for advanced driver assistance and autonomous-driving systems because recognition errors under real road conditions can directly affect navigation, warning, and safety-related decisions. This study presents a reliability-oriented comparison of BaselineCNN, ResNet18, MobileNetV2, MobileNetV2 enhanced with the Convolutional Block Attention Module (CBAM), and knowledge-distilled MobileNetV2. All models were trained from scratch on a fixed stratified split of the German Traffic Sign Recognition Benchmark (GTSRB) using five independent random seeds. The evaluation considered clean classification performance, training stability, bootstrap confidence intervals, McNemar paired tests, probabilistic calibration, severity-wise robustness under blur, central occlusion, low-light, and Gaussian noise corruptions, external validation on 360 cropped German Traffic Sign Detection Benchmark (GTSDB) signs, computational efficiency, and Grad-CAM-based diagnostic analysis. Across five seeds, ResNet18 achieved the strongest mean clean performance, with an accuracy of 0.9856 ± 0.0093 and macro-F1 of 0.9817 ± 0.0134. MobileNetV2 remained competitive, with an accuracy of 0.9813 ± 0.0057 and macro-F1 of 0.9773 ± 0.0069, whereas BaselineCNN was substantially weaker, with an accuracy of 0.8459 ± 0.0165 and macro-F1 of 0.8384 ± 0.0189. ResNet18 also showed strong calibration, with an expected calibration error of 0.0017, and achieved the best GTSDB macro-F1 of 0.9389 in the representative Seed-42 external evaluation. Severe central occlusion was the most damaging corruption, reducing all models below 0.11 macro-F1, while low-light degradation was comparatively less harmful for the stronger classifiers. The results show that model ranking changes across accuracy, calibration, robustness, external transfer, computational cost, and visual diagnostic behavior. Therefore, traffic sign classifiers should be selected using multi-seed, multi-metric evaluation rather than clean benchmark accuracy alone.

Algorithms doi: 10.3390/a19070556

Authors: Manu Chaudhary Kareem El-Araby Devon Bontrager Alvir Nobel Shima Mohaghegh Kieran Egan Manish Singh Trey Campbell Jacob Spry Luis Aviles Naveed Mahmud Pranav Reddy Pruthviraj Sadhankar Shivansh Shrivas Esam El-Araby

Partial differential equations (PDEs) play a central role in scientific and engineering analysis, with applications spanning fluid dynamics, heat and mass transfer, electromagnetism, quantum mechanics, and financial modeling, where they are used to describe diffusion processes, wave propagation, and the evolution of complex systems over space and time. Solving multidimensional partial differential equations (PDEs) is a computationally challenging problem, even for the most advanced classical systems. Over the past decade, quantum computing has attracted significant interest as a potential approach for solving complex computational problems, including multidimensional PDEs. Although a variety of approaches have been proposed for solving PDEs, most of the existing techniques are based on variational quantum algorithms (VQAs). Despite being promising, these VQA-based approaches suffer from low accuracy, long execution times, and limited scalability. In this work, we propose a scalable and efficient quantum algorithm for solving multidimensional PDEs. Our algorithm has two variants. One variant is based on the finite difference method (FDM), classical-to-quantum (C2Q) encoding, and numerical instantiation, whereas the other is based on FDM, C2Q, and column-by-column decomposition (CCD). We have also evaluated our algorithm using several practical case studies; namely, Poisson, heat, Black–Scholes, and Navier–Stokes equations. The results show that our proposed approach achieves higher accuracy, greater scalability, and faster execution time than the VQA-based approaches. We validated these findings on both noise-free and noisy simulators, as well as on a hardware emulator and real IBM quantum hardware.

Algorithms doi: 10.3390/a19070555

Authors: Aurora Gensale Luca Cagliero Cataldo Basile Paolo Garza Luca Ferrua

Hundreds of connected components expose vehicle systems to an increasing number of cyber attacks. Vehicle manufacturers must establish the most appropriate procedures for vulnerability assessment and penetration testing using a mix of proprietary solutions and open-source standards. The spread of Large Language Models (LLMs) simplifies the interaction between automotive experts and domain-specific knowledge bases. While proprietary LLM services can be expensive and raise data privacy concerns, open-source LLMs are potentially more cost-effective and better suited to in-house solutions. However, the effectiveness of open-source models in retrieving automotive-related cybersecurity information remains unclear. While adopting open-source LLMs with a few billion parameters, their reasoning and generative capabilities under in-context learning settings are questionable. To bridge this gap, this paper explores efficient solutions for Retrieval-Augmented Generation (RAG) architecture for automotive cybersecurity relying on open-source LLMs. The ultimate goal is to enable cost-effective retrieval and question answering from in-domain knowledge bases, overcoming the privacy and confidentiality issues raised by automotive experts. Using a Graph Knowledge Base designed for a corporate scenario, this paper first defines an expert-curated testing benchmark to evaluate in-domain question-answering performance across multiple aspects. Next, it proposes different RAG system variants based on various retrieval strategies and LLMs, both proprietary and open-source. Finally, it quantitatively evaluates the effectiveness of the content retrieval strategies and compares the pertinence, conciseness, and completeness of generated answers through human validation. Notably, within the scope of the performed analysis, RAGs that rely on open-source models demonstrate promising and competitive performance in some respects compared to the OpenAI GPT model. RAG retrieval performance also surpasses that of state-of-the-art solutions on existing cybersecurity benchmarks (Recall@K above 0.95 vs. 0.65 for state-of-the-art in-domain RAGs).

Algorithms doi: 10.3390/a19070554

Authors: Gianmarco Baldini Fausto Bonavitacola

Global Navigation Satellites Systems (GNSSs) have become a predominant feature in the digital life of citizens, and they provide positioning services in various applications including pedestrian and vehicular navigation. In urban environments with the presence of buildings and another obstacles, GNSS positioning may be unreliable because of non-line-of-sight (NLOS) conditions, and the classification of observed satellite visibility between LOS and NLOS may improve GNSS receivers to improve their performance to provide the positioning services. In this context, machine learning algorithms using features like signal noise ratio, pseudorange, elevation angle, and others have been applied to this problem both in supervised and unsupervised mode. Because the ground truth information on LOS/NLOS conditions may not always be available, unclustering algorithms have been applied for unsupervised classification, but the classification performance is still limited. This paper proposes an ensemble approach where different clustering algorithms, both historical and recently introduced in the literature, are combined to improve the LOS/NLOS classification accuracy. Even if the ensemble approach manages to achieve a significant improvement, a novel and more sophisticated approach is proposed in this paper, where the contributions of each clustering algorithm are weighted. The optimal values of the weights are estimated using various Meta-Heuristics Algorithms (MHA) on a subset of GNSS data where the ground-truth information is available (i.e., labeled data set). In a subsequent step, the performance of the optimal weighted clustering ensemble is evaluated. The approach is applied to a recent public data set with 57 satellites, where it is shown to outperform the specific clustering approaches by a large margin (more than 7%). The Meta Heuristics Algorithm (MHA)s have similar performance, with the Dynamic Opposition Grey Wolf Optimization (DOLGWO) having a minor advantage against the other MHAs.

Algorithms doi: 10.3390/a19070553

Authors: Brian A. Polin Alexander Rotshtein Denis Katelnikov Oksana Zelinska

We propose a hybrid intelligent methodology for forecasting outcomes in complex human-centered systems characterized by uncertainty and reliance on expert knowledge. The framework integrates fuzzy cognitive maps (FCMs), a novel Least-Influence Method for estimating causal arc weights, and genetic algorithms for model tuning. The proposed influence comparison method simplifies expert elicitation by reducing the cognitive load of direct weight estimation, while the genetic algorithm ensures alignment of forecasts with observed or expert-derived data. A forecasting algorithm based on incremental changes in concept levels enhances the sensitivity of the output variable to factor variations. To illustrate the applicability of the framework, we construct a decision-support model for predicting weight-loss success under diverse psychological, behavioral, and environmental conditions. Simulation results demonstrate how factor ranking, scenario modeling, and paired influence analysis provide actionable insights for decision-making. Beyond the weight-loss domain, the approach is generalizable to a wide range of knowledge-based systems requiring robust integration of expert judgment, fuzzy reasoning, and evolutionary optimization.

Algorithms doi: 10.3390/a19070552

Authors: Baudoin Nguimeya Tsofack Garrik Brel Jagho Mdemaya Milliam Maxime Zekeng Ndadji Maxwell Ndognkom Manga Mthulisi Velempini

The efficient resolution of complete NP problems, such as the Traveling Salesman Problem (TSP), particularly for large instances, remains a major challenge in operations research and combinatorial optimization, especially for many businesses, particularly in sectors such as logistics, urban planning, and networks, where efforts are made daily to optimize routes and delivery times. Optimization methods inspired by collective behavior, such as Ant Colony Optimization (ACO), offer competitive results for solving these types of problems. The main problem is the size of the instances because, when it becomes large, many existing algorithms fail to converge to a good solution within a reasonable timeframe: the execution time is generally very long, and the solution obtained is generally far from being the optimal solution to the problem. In this article, we propose a new way of approaching the resolution of the TSP through new metaheuristics inspired by artificial intelligence techniques and ant colony theory. To evaluate the effectiveness of our methodology, particularly the Multi-colony Ant Colony Optimization version 2-SK (MACOV2SK) method, simulations were performed on several instances of the TSP, focusing on large-scale instances. The experimental results clearly demonstrate that the proposed approach significantly improves upon several other approaches in the literature in terms of execution time and solution quality, especially for large-scale problems.

Algorithms doi: 10.3390/a19070551

Authors: Carlos Fernando Mourão Luiz Eduardo Juliasse Adam Lowenstein Bruno César de Vasconcelos Gurgel Rodrigo dos Santos Pereira Gutemberg Gomes Alves

Randomization sequence generation is essential in randomized controlled trials, but access to trial-management systems or statistical support may be limited in some settings. This in silico technical benchmarking study evaluated whether four LLM-linked chatbot interfaces can faithfully execute pre-specified deterministic Python code to generate a randomized sequence under fixed-seed conditions. In Experiment 1, four investigators performed 1200 fixed-seed Python runs across two sample-size scenarios (n = 30 and n = 50), benchmarked against seeded Excel/VBA and R Console workflows. In Experiment 2, the same investigators performed 320 NL-only runs without code submission or seed specification. A supplementary permuted block benchmark (n = 60; blocks of six) added 640 runs across both prompting conditions. Fixed-seed code execution achieved 100% exact reproducibility, allocation integrity, format compliance, and operational completion across all platforms. NL-only prompting preserved allocation integrity, format compliance, and operational completion (100%) but yielded 0% exact reproducibility in both simple and permuted block randomization. These findings support only a constrained interpretation: chatbot-mediated reproducibility depends on executable code, fixed-seed specification, preserved documentation, and human verification. These interfaces should not replace dedicated randomization software or validated trial-management systems.

Algorithms doi: 10.3390/a19070550

Authors: Abdullah Albanyan Ali Louati Hassen Louati

Traffic signal control increasingly faces disturbed operating conditions such as incidents, abrupt demand surges, sensing degradation, and abnormal driving patterns. Under these nonstationary regimes, classical fixed-time and actuated strategies may exhibit slow recovery, while purely data-driven controllers can be brittle when disturbance characteristics shift. This paper proposes an LLM-in-the-loop architecture for disturbed traffic signal control that integrates (i) deep learning for disturbance detection and short-horizon traffic forecasting, (ii) a disturbance-aware candidate generation and scoring layer (template/retrieval-based), and (iii) a constrained large language model (LLM) that selects or minimally repairs signal plans only within constraint-screened action templates. A deterministic validator enforces safety and operational constraints, including minimum/maximum greens, cycle feasibility, and clearance rules, by checking action feasibility before execution. The method is formulated as constrained decision making under uncertainty, where disturbance estimates and predictive confidence shape both retrieval/scoring and LLM supervision. The originally reported SUMO evaluation considered multiple disturbance categories, including capacity drops, demand shocks, and sensing dropouts as well as reported network delay, queue spillback, recovery time, and switching stability. Within the originally reported SUMO scenarios, descriptive results suggest that among the selected baselines, the proposed DL + LLM framework reported lower mean values of delay, spillback frequency, and recovery time than the fixed-time, actuated, and retrieval-only baselines. The reported validator-detected action-feasibility violations were zero; this result concerns timing-action feasibility rather than an absence of traffic-state risks such as spillback.

Algorithms doi: 10.3390/a19070549

Authors: Dora Ballesteros Daniel Suarez Cesar Pachon

Audio deepfake detection systems often achieve excellent internal validation performance but fail to generalize under real-world inference conditions involving synthetic speech generated with previously unseen AI tools. To address this limitation, this work proposes the Spectral Multi-Representation Fusion (SMRF) framework, which integrates multiple spectral representations and decision-level fusion strategies to improve robustness under cross-domain conditions. Additionally, a Stability-Aware Multi-Metric Selection (SAMMS) strategy is introduced to select architectures by jointly considering predictive performance and cross-representation stability. The proposed framework was evaluated using four spectral representations (log-magnitude spectrogram (LOG), Mel spectrogram (MEL), Discrete Wavelet Transform (DWT), and Constant-Q Transform (CQT)) combined with multiple convolutional architectures and complementary voting strategies. The experiments revealed that isolated models exhibiting validation metrics above 95% may still produce very poor synthetic-audio detection rates during external inference (even lower than 10%). In contrast, fusion-based strategies substantially improved robustness by exploiting complementary synthetic evidence across spectral domains. The results also demonstrated that both the voting strategy and the SAMMS stability parameter λ strongly affect the final behavior of the system. In particular, hybrid fusion using One-Hard Voting with two architectures selected using λ≥0.25 achieved the best balance between synthetic-audio detection and real-audio preservation, outperforming individual models under cross-domain inference conditions, with detection rates close to 75% for both synthetic and real audio. These findings suggest that stability-aware fusion strategies constitute a promising direction for improving robustness in realistic audio deepfake detection scenarios.

Algorithms doi: 10.3390/a19070548

Authors: Vesna Antoska Knights Valbona Mazlami Marija Prchkovska Jasenka Gajdoš Kljusurić

Feature importance analysis is essential for interpreting machine learning models in diabetes mellitus (DM) risk prediction; however, existing interpretability methods often produce inconsistent feature rankings across models. This study proposes a unified ODE-inspired interpretability framework and an algorithmic decision procedure for robust feature selection by integrating contribution-based (SHAP), perturbation-based (permutation importance), and sensitivity-based feature importance measures. Multiple supervised machine learning models, including Logistic Regression, Random Forest, Gradient Boosting, Histogram Gradient Boosting, and Multilayer Perceptron, were trained and evaluated on a longitudinal biochemical and demographic dataset comprising 200 patients with three repeated visits (N = 600 observations). To preserve longitudinal integrity and avoid patient-level information leakage, grouped cross-validation was applied. A sensitivity-based feature importance formulation using finite-difference approximations enabled model-agnostic comparison across heterogeneous machine learning architectures. Stability, normalization, and cross-method agreement analyses were additionally introduced to evaluate consistency of feature rankings across models and interpretability methods. Experimental results consistently identified HbA1c as the dominant predictor, followed by lipid-related variables, age, and body mass index. Strong agreement was observed between ODE-inspired feature importance and SHAP analysis, whereas permutation importance demonstrated comparatively weaker agreement with sensitivity-based methods. The proposed framework further enabled systematic analysis of ranking stability, cross-method agreement, longitudinal sensitivity dynamics, and the introduction of an agreement-weighted Consensus Interpretability Score (CIS) for unified feature ranking across heterogeneous interpretability methods. The results demonstrate that integrating ODE-inspired sensitivity analysis with machine learning provides a robust, interpretable, and computationally scalable framework for feature importance assessment in diabetes risk prediction. The proposed approach offers a principled solution to inconsistent feature importance estimation and supports more reliable interpretation of biomedical machine learning models.

Algorithms doi: 10.3390/a19070547

Authors: Alireza Panahi Salim Lahmiri

Background: Strategic resource planning is crucial for optimizing supply chain management and ensuring efficient operations. This study aims to enhance strategic planning in gold mines by leveraging advanced gold price forecasting models. By predicting future gold prices accurately, mining companies can better plan their extraction, processing, and distribution activities, thereby improving overall supply chain efficiency. Methods: Various advanced forecasting models are implemented, including backpropagation neural networks (BPNNs), convolutional neural network (CNN), long short-term memory (LSTM), bi-directional LSTM (Bi-LSTM), gated recurrent unit (GRU), and bi-directional GRU (Bi-GRU). The feature selection process is facilitated by recursive feature elimination (RFE), and Optuna is used to fine-tune neural network models. Evaluation is based on root mean squared error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE). Results: BPNN performed the best in terms of lowest RMSE (0.5928), MAE (0.4091), and MAPE (0.34%), whilst Bi-GRU was the poorest performer, as it achieved RMSE of 9.41, MAE of 8.1916, and MAPE of 6.94%. In addition, Optuna further improved each model’s accuracy, except CNN, where the performance slightly decreased. Conclusions: Advanced forecasting neural systems underperformed the standard backpropagation neural networks. In this regard, BPNN proved to be highly effective in forecasting gold price, providing critical managerial implications for navigating the dynamic and volatile gold market for gold mining companies and investors.

Algorithms doi: 10.3390/a19070546

Authors: Chingiz Alimbayev Zhadyra Alimbayeva Kassymbek Ozhikenov Kairat Karibayev Zhanat Abuova Dilfuza Akhmedova

Diabetes mellitus is strongly associated with cardiovascular dysfunction and remains one of the leading contributors to morbidity and mortality worldwide. Early identification of diabetes-related cardiovascular alterations is essential for timely risk stratification and personalized clinical management. In the present study, an interpretable machine learning framework for diabetes classification in patients with cardiovascular disease was developed using routinely available clinical, biochemical, renal, and echocardiographic parameters. A retrospective dataset consisting of 131 cardiovascular patients was included in the final analysis, comprising 65 patients with diabetes mellitus and 66 patients without diabetes. Demographic, metabolic, renal, and cardiovascular variables, including age, body mass index (BMI), glycated hemoglobin (HbA1c), glucose concentration, estimated glomerular filtration rate (eGFR), troponin level, heart rate, and left ventricular ejection fraction (EF), were included in the analysis. Multiple supervised machine learning algorithms, including Logistic Regression, Support Vector Machine (SVM), Gradient Boosting, and Random Forest, were implemented and compared using repeated stratified cross-validation. Among the evaluated models, Random Forest demonstrated the highest classification performance, achieving a mean ROC AUC of 0.880 ± 0.050. Statistical analysis revealed significantly elevated HbA1c, glucose, and troponin levels together with reduced ejection fraction values in diabetic patients. Explainable artificial intelligence analysis using SHAP and partial dependence plots identified glucose concentration, HbA1c, age, and renal function as the dominant contributors to diabetes classification. Nonlinear relationships between metabolic and cardiovascular variables were additionally observed. The obtained findings demonstrate that interpretable machine learning approaches can provide effective discrimination between diabetic and non-diabetic cardiovascular patients while maintaining clinically meaningful interpretability. The proposed framework may contribute to future intelligent clinical decision-support systems and personalized cardiovascular risk assessment strategies.

Algorithms doi: 10.3390/a19070545

Authors: Salah Hanafi Mohammed-Karim Fellah Youcef Djeriri Habib Benbouhenni Abdelkder Achar Mohamed Fouad Benkhoris Patrice Wira Nicu Bizon

This paper presents an adaptive neural-network-based algorithm for chattering mitigation in sliding mode control (SMC) of parallel multicellular DC/AC power converters. Although conventional SMC provides strong robustness against parameter uncertainties, external disturbances, and load variations, its discontinuous control action often generates chattering, resulting in excessive switching activity and reduced converter performance. To address this limitation, a computationally efficient adaptive neural network is integrated into the SMC framework to approximate the discontinuous switching term and generate a smooth control signal. The proposed algorithm updates neural network parameters online through an adaptive learning mechanism, enabling real-time compensation of modeling uncertainties while preserving the inherent robustness of SMC. The resulting adaptive neural network sliding mode control (ANN-SMC) algorithm is formulated to ensure accurate output voltage tracking, balanced operation of converter cells, and reduced switching oscillations. Extensive simulation studies are conducted under different operating scenarios, including load variations and system disturbances. The performance of the proposed method is evaluated against classical SMC using quantitative indicators related to tracking accuracy, dynamic response, robustness, and chattering suppression. The results demonstrate that the ANN-SMC algorithm significantly reduces high-frequency oscillations while improving transient behavior and maintaining robust operation. These findings indicate that the proposed adaptive learning-based control algorithm constitutes an effective and scalable solution for advanced power conversion systems operating under uncertain conditions.

Algorithms doi: 10.3390/a19070544

Authors: Ong Andre Wahyu Riyanto Budi Santosa Nurhadi Siswanto

Optimized wind turbine configurations enable sustainable power generation; however, maximizing total power output through wind farm layout optimization (WFLO) remains a critical and complex challenge. WFLO strategically identifies the optimal positioning of the wind turbines within a designated wind farm area to mitigate wake effects and maximize overall power output. Wake effects, where upstream turbines diminish the wind velocity reaching downstream turbines, can substantially lower the overall power output of a wind farm. This study proposes an ant colony optimization (ACO) approach to address WFLO, employing a continuous wind farm model that enables flexible turbine placement to minimize the cost of energy (CoE). Three scenarios were examined: constant wind speed in a single wind direction; constant wind speed in multiple wind directions; and a variable wind speed in multiple directions incorporating direction-specific occurrence probabilities. The proposed framework facilitates continuous spatial optimization, substantially expanding the feasible search space by eliminating grid-based constraints. Its effectiveness was validated through comparative evaluation against methods reported in prior studies. The results demonstrate that the ACO algorithm consistently achieves superior performance, yielding a 2.13–4.03% reduction in CoE and a 2.22–4.06% increase in total power output across diverse wind conditions. This study highlights the potential of continuous spatial optimization as an effective alternative to conventional grid-based approaches in WFLO.

Algorithms doi: 10.3390/a19070543

Authors: Hongjin Huang Xia Geng

Visible-infrared person re-identification (VI-ReID) aims to retrieve pedestrian images of the same identity across visible and infrared modalities, but remains challenging due to the large modality gap and unstable local correspondence. Existing methods mainly rely on visual cues, which may be insufficient when infrared images lack color and fine-grained texture information. To address this issue, this paper proposes a CLIP-Guided Progressive Body-Part Semantic Alignment Network, termed PBSA-Net. The proposed method introduces CLIP-derived textual semantics as modality-agnostic guidance for both global representation learning and local body-part feature extraction. Specifically, a global semantic branch first learns identity-level textual anchors to regularize global visual features. Then, a body-part semantic branch exploits identity-aware body-part prompt learning, multi-level feature fusion, and text-guided cross-attention to guide fine-grained local representation learning. A progressive three-stage optimization strategy is further adopted to decouple global semantic learning, body-part semantic correspondence learning, and retrieval-oriented feature optimization. Experiments on SYSU-MM01, RegDB, and LLCM demonstrate the effectiveness of PBSA-Net. It achieves 76.5% Rank-1 and 74.2% mAP on SYSU-MM01, 82.5% Rank-1 and 76.0% mAP on RegDB, and 61.8% Rank-1 and 65.8% mAP on LLCM. Ablation studies further show that the proposed body-part semantic alignment and progressive optimization provide complementary improvements.

Algorithms doi: 10.3390/a19070541

Authors: Stefan Pitulić Dragana Radosavljević Đurica Marković Siniša Ilić

This paper presents a hybrid algorithmic framework for nonparametric multisite daily streamflow simulation, evaluated on 52 years of observed data from three hydrological stations. The method generates streamflow data of 1000 synthetic years while preserving marginal distributions, daily rank structures, inter-station consistency, annual-sum behavior, and dependence across consecutive years. The workflow integrates Monte Carlo sampling, Schaake-shuffle reordering, block-mosaic reconstruction with partial freezing, Hungarian assignment optimization, annual-sum matching, and an adaptive permutation genetic algorithm for year-order optimization. The results show that the proposed algorithm improves aggregate hydrological diagnostics, particularly annual-sum autocorrelation, hydrological indices, persistence, seasonality, and timing of extremes, while reducing runtime in the final optimization phase by 45.2% compared to the benchmark algorithm. The study therefore formulates daily streamflow simulation as a constrained time-series reconstruction and permutation-optimization problem, making the method suitable for further algorithmic development and other multisite environmental time-series applications.

Algorithms doi: 10.3390/a19070542

Authors: Ntebogang Dinah Moroke

We propose a family of spectral hypergraph algorithms for early detection of connectivity collapse in pharmaceutical supply chain networks. The Fiedler eigenvalue λ2 of the normalised hypergraph Laplacian serves as the order parameter. Five geometry-aware early warning indicators (TSI, HSST, HOMFA, HOTV, ORC) monitor network topology rather than scalar residuals, with provable detection guarantees under geometric ergodicity. A Greedy Dejamming algorithm restores connectivity via rank-2 Laplacian updates, achieving a (1 − 1/e)-approximation within a procurement budget constraint. Monte Carlo validation on a calibrated pharmaceutical distribution hypergraph demonstrates substantially higher detection sensitivity and shorter lead times than classical statistical process control. Hyperedge representation yields detection gains exceeding 90% for simultaneous multi-party failures that pairwise graph projections miss entirely. A COVID-19 lockdown episode provides a held-out directional consistency check.

Algorithms doi: 10.3390/a19070540

Authors: Limei Zhang Jiaman Li Yuhan Song Shuang Wang Weijie Dong

In smart grid operation, the scarcity of abnormal samples causes data imbalance, which is a key factor limiting the accuracy of anomaly detection. To address this issue and simultaneously solve the problem of easily losing weak abnormal signals in one-dimensional time-series data, an abnormal data detection method for grid information using an improved AlexNet considering sample imbalance is proposed. Firstly, features like voltage, current, and power are extracted from historical data. Missing values are filled via Lagrange interpolation, and abnormal boundaries are determined using box plots to construct high-quality samples. Secondly, to address the problem of few abnormal samples and imbalanced distribution, an enhanced learning strategy combining time-series translation and Gaussian noise injection is adopted to expand the abnormal samples and obtain sufficient training data. Then, to preserve the integrity of weak signals in one-dimensional time-series data and amplify the differences in abnormal features, the Gram angle field is used to convert multi-dimensional time-series data into a two-dimensional image, achieving the visual representation of time-series features. Finally, combined with the powerful image detection capability of AlexNet, it is improved by lightweighting the network structure, introducing the multi-head self-attention mechanism, and optimizing the training strategy to adapt to abnormal detection in the small sample and imbalanced environment of the grid. The simulation experiments show that the proposed method achieves an accuracy rate of 91.32% on extremely imbalanced datasets, which is at least 3.1% higher than those of other models.

Algorithms doi: 10.3390/a19070539

Authors: Qi Li Xiaolei Wang Shuo Wen Wei Du Li Mao Lijun Cai

Approximation jobs are widely deployed on Amazon EC2, which compute partial task segments to obtain useful results. For such jobs, maximizing total profit is the primary goal, where profit equals the sum of job utilities minus the total machine costs. Unfortunately, maximizing the total profit of approximation jobs is an NP-hard problem. This problem is further complicated by online job arrivals and heterogeneous resource demands across different tasks. This work builds an optimization framework that clearly characterizes job utility and machine costs to resolve this problem. Within this framework, we propose an efficient dual algorithm for job scheduling. The proposed method leverages the dual-fitting approach to measure algorithm performance by analyzing the primal and dual objective growth at each step. This work proves that our algorithm achieves a constant competitive ratio. The results from the trace-driven simulations demonstrate that our algorithms consistently outperform these baselines across various metrics.

Algorithms doi: 10.3390/a19070538

Authors: Youpeng Sun Peng Li Wenhui Wang Ye Xu Wenqi Geng Jiajun Ding

The Poisson multi-Bernoulli mixture (PMBM) filter is an effective approach for multi-object tracking in complex scenarios. However, its performance deteriorates when surviving objects spawn, as the PMBM filter only classifies detected objects as either new-born or surviving, thereby ignoring information from the surviving objects and preventing timely identification of spawning events. To address this limitation, this paper proposes the Dynamic Hybrid Detection-Gamma Gaussian inverse Wishart Spawning Poisson multi-Bernoulli mixture (DHD-GGIW-SPMBM) filter, which models spawning objects independently using a Bernoulli process to enhance tracking accuracy. The probability generating functional is employed to derive the recursive prediction and update equations of the proposed filter, and its conjugacy after prediction and update is formally proven. Additionally, a dynamic hybrid detection method is introduced to evaluate the consistency between measurements and theoretical samples, enabling the detection of spawning events. The detection results guide an evidential Gaussian mixture model (EGMM) for fuzzy partitioning of the spawning process, reducing errors under closely spaced and high-clutter conditions. Simulation results demonstrate that, compared with existing spawning-capable filters, the proposed DHD-GGIW-SPMBM filter achieves superior tracking performance, faster identification of spawned objects, and robust operation in complex scenarios.

Algorithms doi: 10.3390/a19070537

Authors: Mantong Huang Xin Yu Rixin Lin

This paper proposes an adaptive single-layer distributed neurodynamic optimization approach with the penalty method to address a non-smooth pseudoconvex optimization problem with affine equality and inequality constraints in multi-agent systems, where the global objective function for the agents is pseudoconvex but not required to be differentiable. The target of this approach is to optimize the global objective while ensuring compliance with various constraints. The approach avoids the use of additional auxiliary variables, thereby reducing communication bandwidth and computational complexity. Under mild assumptions, the solution of the designed model is bounded for any initial conditions, to enter their respective feasible domains in finite time, and remain within these domains indefinitely. To achieve finite-time consensus in undirected, connected networks for multi-agent systems, a novel consensus mechanism is introduced to ensure that all agents synchronize their states within finite time. By exploiting the unique pseudoconvexity of the global objective function, the solution trajectory converges to the optimal state of the original problem. Furthermore, the effectiveness of the proposed approach is verified through two simulation experiments, and comparisons with four existing algorithms are conducted to demonstrate its superiority in convergence performance. Finally, an economic dispatch problem in power systems is provided as an engineering application to illustrate the practical applicability of the proposed algorithm.

Algorithms doi: 10.3390/a19070536

Authors: Yingjie Xie Yinbo Liu Yanfei Liu Junfang Li Wenjun Wang

Change point detection in dynamic social graphs aims to identify significant transitions in evolving interaction patterns. The task is particularly challenging due to sparse and noisy interactions and frequent user turnover, while the existing methods largely overlook the semantic and emotional signals in user-generated content by focusing primarily on structural changes. To address these limitations, this paper proposes ECPD-SG, an emotion-aware contrastive prototype learning algorithm for unsupervised change point detection in dynamic social graphs. ECPD-SG constructs emotion-aware graph snapshots by integrating textual and affective features into node representations and recalibrating interaction weights through emotion-aware attention. It then summarizes temporal node representations into adaptive prototypes and models their evolution using optimal-transport-based alignment and contrastive learning. Change points are detected from prototype-level shift scores with an adaptive CUSUM decision rule. Experiments on real-world dynamic social graph datasets show that ECPD-SG achieves competitive or superior performance over representative baselines, while ablation and sensitivity analyses verify the effectiveness of its key components.

Algorithms doi: 10.3390/a19070534

Authors: Daniel Mejía Héctor López Leonel Estrada Yann E. Bouvier Joaquín Vaquero Nimrod Vazquez José Magaña

This paper presents a Finite Control Set-Model Predictive Control (FCS-MPC) strategy for a modified single-phase 17-level Double Flying-Capacitor Multilevel (DFCM) converter. The proposed approach integrates current regulation, capacitor voltage balancing, switching frequency reduction, delay compensation, and FPGA-based real-time implementation within a unified predictive control framework. A multi-objective cost function exploits the converter’s redundant switching states to achieve accurate control while reducing computational burden. Additionally, the converter topology provides voltage-boosting capability without requiring an additional DC-DC stage. The proposed controller was validated through offline and Hardware-in-the-Loop (HIL) simulations. Simulation results demonstrate robust operation, effective capacitor voltage balancing, and excellent current quality, achieving a THD of 0.7%.

Algorithms doi: 10.3390/a19070535

Authors: Kunchang Wu Xiaomin Wu Hua Shi Miao He

Accurate remaining useful life (RUL) prediction of lithium-ion batteries is critical for the safe and reliable operation of battery management systems. While point prediction methods have been extensively studied, principled uncertainty quantification (UQ) remains underexplored, particularly in the early-cycle regime where degradation signals are subtle and training data are scarce. This paper presents a systematic evaluation of ten UQ configurations for early-cycle battery RUL prediction using the MIT-Stanford dataset of 124 cells. A composite-kernel Gaussian process regression model combining a radial basis function and a white noise kernel, denoted GPR-CK(RBF + W), is used as the core predictor. We compare Bayesian native UQ, split conformal prediction, jackknife+, bootstrap resampling, and a Conformalized Adaptive Intervals (CAI) method across a Primary test set and a distribution-shifted Secondary test set collected one year later. On the Primary test set, the composite-kernel GPR variants achieve full (100%) prediction-interval coverage at the 95% nominal level, while the proposed CAI calibration yields the sharpest coverage-valid intervals. Under a one-year distribution shift, GPR-CK(RBF + W) empirically retains 97.5% coverage, which we report as empirical robustness rather than a guaranteed coverage level. A leave-one-out calibration factor (δ = 1.39 with the white noise kernel versus δ = 3.43 without it) isolates explicit noise modeling as the decisive factor for calibration. Feature dimensionality analysis further reveals a three-phase sensitivity pattern, identifying three features as the optimal operating point balancing predictive accuracy and UQ quality.

Algorithms doi: 10.3390/a19070533

Authors: Luis Angel Iturralde Carrera Leonel Díaz-Tato Guillermo José Barroso García Yoisdel Castillo Alvarez Yarelis Valdivia Nodal Miguel Angel Cruz-Pérez Juvenal Rodríguez-Reséndiz

The transition toward next-generation energy systems requires advanced computational tools capable of supporting accurate, adaptive, and data-driven energy performance assessment. Within this context, Energy Performance Indicators (EnPIs) established under the ISO 50001 framework remain essential for monitoring energy efficiency and continuous improvement; however, conventional indicators are often based on static or simplified relationships that do not adequately capture the dynamic, nonlinear, and multivariable behavior of modern buildings and energy management systems. This systematic review analyzes the integration of ISO 50001-based EnPIs with intelligent algorithms and artificial intelligence techniques for enhanced energy management. The review follows a PRISMA-inspired methodology, using Scopus as the primary database and Web of Science and Google Scholar as complementary sources. From 5442 initial records, 2691 studies were screened and 283 articles were selected for detailed analysis, supported by a bibliometric keyword co-occurrence analysis using VOSviewer 1.6.20. The results show a clear evolution from traditional energy indicators and normalized baselines toward computational modeling approaches based on regression analysis, machine learning, deep learning, forecasting, anomaly detection, and optimization algorithms. These methods improve the predictive capability, adaptability, and operational relevance of EnPIs by incorporating climatic, occupancy, temporal, and operational variables. The reviewed evidence indicates that intelligent algorithms can strengthen ISO 50001 energy management systems by enabling dynamic baselines, early detection of abnormal consumption patterns, predictive decision-making, and continuous operational optimization. Nevertheless, challenges remain regarding data quality, model interpretability, methodological standardization, and practical integration into certified energy management frameworks. Overall, this review highlights that the future of energy performance assessment does not rely on replacing conventional EnPIs, but on transforming them into intelligent, computationally supported indicators for sustainable, resilient, and next-generation energy management systems.

Algorithms doi: 10.3390/a19070532

Authors: Zhaoguang Zhang Zhenhua Wei Siming Han Yuxin Yang Jianwei Zhan Wenpeng Wu Haiyang You Chenxi Li

Automatic modulation recognition (AMR) is a key technique in modern communications. We propose LM-GDMAF, which integrates lightweight Mamba modules with a GDMAF fusion module. The lightweight Mamba modules efficiently extract temporal features, while GDMAF deeply fuses dual-modality information from IQ time series and frequency-domain spectrograms. On a public dataset, the proposed method improves average recognition accuracy by at least 4.25% compared to baseline methods. The number of computational parameters of the lightweight Mamba module is reduced by approximately 26.1% compared to the standard Mamba model. The complete model achieves average modulation recognition accuracy gains of approximately 11.89% and 14.74% over the respective single-branch models.

Algorithms doi: 10.3390/a19070531

Authors: Ilaria Ciampa Benedetta Perrone Umberto Mosca Elisa Fattori Serena Sinagra Alessandro Cicolin Irene Rechichi Gabriella Olmo

Polysomnography (PSG) is the gold standard for diagnosing sleep disorders, but its complexity and cost limit widespread use. Heart rate variability (HRV) is traditionally assessed from electrocardiography (ECG), while photoplethysmography (PPG), widely available in wearable devices, offers a more accessible alternative. However, its reliability over full-night recordings remains underexplored. This study analyzes data from 50 subjects across five groups (healthy controls, rapid eye movement sleep behavior disorder, obstructive sleep apnea, periodic limb movements, and mixed comorbidities) to assess agreement between ECG-derived HRV and PPG-derived pulse rate variability (PRV), considering time-, frequency-, and nonlinear-domain features. Correlation and equivalence analyses were performed, with and without removal of artifactual segments. Correlation coefficients exceeded 0.6 for most features and improved to above 0.7 after artifact removal. Consistent improvements were observed across all subject groups. Equivalence testing further identified a subset of features showing high agreement and low bias. The results indicate that, with appropriate pre-processing, PPG can approximate ECG-derived variability in full-night sleep recordings. The identification of robust features for screening purposes supports the use of PRV for wearable-based screening and monitoring in heterogeneous sleep disorder populations.

Algorithms doi: 10.3390/a19070530

Authors: Zobeir Raisi

Distinguishing authentic gold products from high-quality counterfeits is a challenging fine-grained computer vision problem; counterfeit items are engineered to replicate surface texture, hallmark engravings, color, and geometry with remarkable fidelity, making visual discrimination unreliable even for trained professionals. In this paper, we address the problem of visual gold authentication from unconstrained smartphone imagery in three main contributions. First, we introduce GoldNet, a public benchmark dataset designed for this task, comprising 2127 real-world images of authentic and counterfeit gold items collected under diverse real-world conditions. Second, we evaluate fourteen classification architectures spanning classical handcrafted texture descriptors, convolutional neural networks (CNNs), and vision transformers under a rigorous transfer learning protocol, establishing the first comprehensive baseline for this problem. Third, we propose GoldFormer, a hybrid dual-stream algorithm that combines the local texture representations of ResNet-50 with the global contextual modeling capability of the Swin Transformer (Swin-T) through a newly designed Texture-Aware Attention Gate (TAAG) module. The TAAG dynamically modulates Swin feature dimensions using CNN-derived texture energy, providing improved discriminability and per-prediction interpretability without requiring post hoc attribution. Experimental results show that, under matched-resolution 5-fold cross-validation, the proposed GoldFormer attains the highest overall accuracy (95.02%, F1-score 0.9502) at roughly half the FLOPs of its higher-resolution setting, statistically tied with the strongest individual backbone (ViT-B/16, 94.31%; McNemar p=0.23) and on par with a training-free soft-voting ensemble (94.92%), while significantly improving on its own Swin-T backbone (93.65%) and adding built-in, attribution-free texture-gate interpretability. GoldFormer surpasses trained human-expert performance (89.80%) by approximately 5 percentage points.

Algorithms doi: 10.3390/a19070529

Authors: Kushal Guha Bakshi Sagnik Sinha Ramakant Bhardwaj Purvee Bhardwaj Satyendra Narayan

In this article, we study semi-Markov decision processes (SMDPs) under the limiting ratio average (undiscounted) pay-off criterion, where the state space is finite and the action space of the decision maker is possibly countably infinite. We impose no restriction on the reward function. We prove that the value of such an SMDP exists and that the decision maker possesses a near-optimal (ϵN-optimal) deterministic (pure) semi-stationary strategy, and we give a truncation-based algorithm that computes the value and such a strategy in finite time by solving a finite linear program at each truncation level. We analyze the algorithm’s computational complexity, showing that each truncation level is solved in time polynomial in the number of state–action pairs, while the number of levels required for termination depends on the prescribed accuracy and on the model. We further establish a certified, computable bound on the truncation error under bounded rewards together with standard ergodicity and tail-regularity conditions, yielding a stopping rule that guarantees a prescribed accuracy; we delimit precisely the regime in which such a certificate exists. Numerical experiments, including randomly generated instances, illustrate the algorithm and its computational behavior. Finally, under standard ergodicity conditions, we derive an optimality equation for these SMDP models.

Algorithms doi: 10.3390/a19070528

Authors: Jiahui Na Wensheng Wang

To address the prevalent issues of slow convergence, low exploration efficiency, and large value estimation bias in traditional Deep Q-Networks for intelligent vehicle path planning, this paper proposes an improved Dueling Double Deep Q-Network (D3QN) path-planning method guided by the Dynamic Window Approach (DWA) heuristic. The Dueling Double DQN architecture decouples state value and action advantage representations, while the dual estimator of Double DQN mitigates Q-value overestimation. A Prioritized Experience Replay (PER) mechanism samples transitions non-uniformly based on Temporal Difference error with importance sampling correction, improving the reuse of critical samples and training stability. DWA evaluation criteria are transformed into dense heuristic reward signals, enabling the agent to receive continuous multi-dimensional guidance during exploration without executing online trajectory optimization. The environment augments the sparse navigation objective with a Chebyshev goal-progress term motivated by potential-based reward shaping theory together with auxiliary DWA-style channels. The policy-invariance property of potential-based shaping is referenced only for the goal term added to the sparse task reward rather than for the full composite training return. A continuous Ackermann steering kinematic model with a pure-pursuit path-tracking controller is adopted for deployment to ensure executable trajectories under non-holonomic constraints. The proposed method (DWA-D3QN) is systematically evaluated against sparse-reward D3QN, PBRS-guided D3QN, DQN, DDQN, Dueling DQN, APF-DQN, PPO, SAC, TD3, A*, and classical DWA in a grid map environment with static and dynamic obstacles. Results are reported with statistical significance over multiple random seeds. Under complex difficulty, DWA-D3QN achieves a success rate of 94.1 ± 3.4% with a collision rate of 5.9 ± 3.4% over 15 seeds, representing improvements of 64.1 and 8.4 percentage points over the sparse-reward and PBRS-guided D3QN baselines, respectively. Ablation experiments reveal the differentiated contributions of clearance, heading, and velocity shaping terms: clearance awareness provides the strongest single contribution, heading alignment reinforces directional guidance, and velocity regularization refines trajectory quality under the joint constraints of the former two. The full composite reward achieves the lowest variance among all evaluated DRL methods, confirming enhanced training stability. Comparisons with PPO, SAC, and TD3 confirm the statistically significant advantages of the proposed framework (PPO: p=0.0010, SAC: p=0.0007, TD3: p=0.0024). ROS/Gazebo validation with an Ackermann-steered vehicle achieves a success rate of 96.0% with a collision rate of 4.0% over 50 trials, further confirming the applicability of the learned policy in continuous-state environments with realistic vehicle kinematics.

Algorithms doi: 10.3390/a19070527

Authors: Shahid Alam Ehab Alnfrawy Amina Jameel Sana Qadir Zahida Parveen Basirah Noor Anushya Arol Imran Chaudhry

The increasing complexity and scale of cyber threats demand intelligent and adaptive defense mechanisms that extend beyond traditional approaches. Artificial Intelligence (AI) has emerged as a key enabler for enhancing cyber security through automated detection, analysis, and response. This paper presents a comprehensive survey of AI applications in cyber security across five major domains: malware detection, intrusion detection, phishing and spam detection, botnet detection, and cyber forensics. A systematic methodology based on data and methodological triangulation is employed to analyze 75 studies published between 2021 and 2025. The paper introduces a multi-layer taxonomy that maps cyber threats to application domains, analysis methods, AI approaches, and their associated capabilities and limitations. In addition, a cross-domain meta-analysis is conducted to identify recurring trends and assess the adoption of AI across different cyber security scenarios. The analysis reveals that deep learning and transformer-based models dominate data-intensive domains such as intrusion detection and malware analysis, whereas traditional machine learning techniques remain effective in structured and resource-constrained settings, particularly for phishing detection. Key challenges include dataset limitations, limited explainability, adversarial vulnerabilities, and computational constraints. Unlike existing surveys that focus on specific techniques or individual cyber security domains, this work provides a unified, application-oriented perspective on AI-driven cyber security. It further highlights emerging trends, open challenges, and future research directions toward more robust, scalable, and trustworthy cyber security systems.

Algorithms doi: 10.3390/a19070526

Authors: Li Liu Chenhao Gu Kaifeng Geng

This paper studies a bi-objective distributed flexible job shop scheduling problem considering machine on–off decisions. A mathematical model is formulated to minimize the makespan and total energy consumption while distinguishing processing energy, idle energy, and on–off energy. To address the coupled effects among job-to-factory assignment, machine selection, operation sequencing, and machine on–off states, a domain-knowledge-driven memetic algorithm (DKMA) is proposed. The algorithm represents each schedule with a three-layer encoding scheme and integrates hybrid initialization, knowledge-driven neighborhood search, and energy-saving reconstruction to improve solution-set quality and the use of on–off-eligible idle intervals. The proposed model and algorithm are evaluated through Taguchi parameter tuning, small-scale mixed-integer linear programming (MILP) validation, component ablation experiments, and multi-algorithm comparisons. The results show that DKMA improves solution-set coverage, Pareto-front approximation, and energy control on the tested instances, which supports its applicability to distributed green scheduling with machine on–off decisions.

Algorithms doi: 10.3390/a19070525

Authors: Zhitao Wu Zonglin Yang Jie Zhou Xuechen Li Hongmin Wang

Capacitated vehicle routing with time windows (CVRPTW) is a natural target for coherent Ising machines (CIMs), but a direct multi-vehicle arc encoding scales as O(mN2) and exceeds the variable budget of current CIM-compatible systems. We argue the bottleneck is encoding density, not expressiveness, and present LSQ, a hardware-aware sparse Quadratic Unconstrained Binary Optimization (QUBO) framework that decouples CVRPTW into a compact customer-to-route assignment QUBO and a classical intra-route ordering step under a soft no-wait service convention. LKH candidate edges compress the per-route edge space from O(N2) to O(KN), and a per-route dynamic-penalty subroutine encodes time-window sensitivities as binary variables in a round-wise outer loop. On a six-vehicle, 51-node reference instance curated from long-term operational data, LSQ shrinks the maximum single-submission QUBO from 15,300 arc variables to 342 logicalQUBOvariables (∼45× compression), cuts travel time by 22.9% (74 vs. 96), and cuts route duration by 11.2% (174 vs. 196) against an OR-Tools soft-window baseline at the same fleet size. OR-Tools retains an advantage on raw time-window penalty (1600 vs. 3540) and runtime; under the scalar operational cost ∑kT(πk)+∑iℓi(τi), OR-Tools is therefore the better single-objective solver, and the comparison is a multi-objective trade-off rather than a scalar dominance claim. Ablations confirm that the LKH prior recovers Held–Karp on a 15-customer TSP at 53 vs. 120 variables and that the dynamic-penalty encoding reduces compressed time-window loss by 15.25% at constant travel. All hardware claims refer to QUBO sizing on a Kaiwu/CIM-compatible backend, not physical CIM execution.

Algorithms doi: 10.3390/a19070524

Authors: Dorota Bartusik-Aebisher Sara Czech Jakub Szpara Avijit Paul Marvin Xavierselvan David Aebisher

Breast cancer remains one of the most significant challenges in modern oncology, while advances in artificial intelligence (AI) are creating new opportunities to improve diagnosis, prognosis, and treatment personalization. The aim of this review was to summarize current and emerging applications of AI in the comprehensive care of patients with breast cancer. This study was conducted as a structured narrative review with elements of integrative evidence synthesis based on publications retrieved from PubMed/MEDLINE, Scopus, Web of Science, and Embase. The review included studies evaluating machine learning and deep learning approaches, such as support vector machines, random forests, convolutional neural networks, Vision Transformers, foundation models, self-supervised learning, federated learning, and multimodal AI systems. The strongest clinical evidence currently concerns AI-supported mammographic screening, where large prospective and real-world studies suggest improvements in cancer detection and workflow efficiency. Applications involving MRI, ultrasound, histopathology, molecular prediction, treatment-response assessment, and treatment selection have shown promising performance, but most remain investigational because of limited prospective multicenter validation. Emerging approaches integrating imaging, pathological, molecular, and clinical data show considerable potential for precision oncology. AI may also support treatment selection, patient monitoring, and survivorship care. Despite promising results, widespread clinical implementation remains limited by challenges related to data heterogeneity, model interpretability, external validation, and integration into clinical workflows. Further prospective multicenter studies are required to establish the safety, reliability, and clinical utility of AI-driven systems in breast cancer care.

Algorithms doi: 10.3390/a19070523

Authors: Andrey Moshkin Maxim Fedorov Vladimir Arlazarov Valeria Gribova Anton Nazarenko Dmitry Repin Olga Klevtsova Aleksandr Romanov

Artificial intelligence (AI) technologies, which are being actively developed in modern medicine today, increase the speed and quality of patient care. This article mainly seeks to demonstrate the use of various options of computer analysis of clinical images to solve practical problems of increasing the efficiency of routine diagnostics using retrospective analysis, as well as show the potential for its widespread implementation (due to the scalability of the architecture) in practical healthcare, exemplified by ultrasound (US) and magnetic resonance imaging (MRI) data analysis. This is an interuniversity study, its research protocol was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the local Ethics Committee of Orel State University named after I. S. Turgenev (Protocol No. 25 dated 16 November 2022). The software was developed using Python 3.7 and open neural network models. Statistical processing included an efficiency assessment for which IBM SPSS Statistics 20.0 was used. Detection errors in the analysis of 550 US cases did not exceed 6–8% and were associated with technical difficulties due to image quality. When studying 1030 MRI studies, only 0.19% of cases failed to obtain reliable image analysis results. The differences in the average values for the dimensional characteristics of the studied vessels were 0.11–0.12 mm. The effectiveness of AI in clinical tasks is presented. The improvement in segmentation accuracy was achieved through the use of step-by-step image optimization during the AI training stage. The evolution of technologies in medicine, aimed at digitalization and personalization, is intended to improve the quality and speed of studying images in practical work.

Algorithms doi: 10.3390/a19070522

Authors: Jimmy Leonardo Rosales Ventocilla Jimmy Aurelio Rosales Huamani Juan Francisco Agreda Vega Evergisto Sare Lara Jose Luis Castillo Sequera Jose Manuel Gomez Pulido

Ultraviolet (UV) radiation is a major environmental risk associated with skin cancer, premature skin aging, and ocular damage. In the context of climate variability, changes in cloud cover and ozone-layer dynamics increase the need for reliable short-term UV forecasting systems in highly exposed urban areas. This study proposes a comparative forecasting framework for UV radiation intensity in Lima, Peru, using more than 827,000 records from a meteorological station. Statistical models, recurrent deep learning architectures, and hybrid neural models were evaluated under a unified protocol including 5 min aggregation, daytime filtering, a fixed 60 min forecasting horizon, chronological train–test partitioning, temporal cross-validation, statistical significance testing, and quantitative residual diagnostics. The results show that recurrent and hybrid deep learning models substantially outperformed traditional statistical approaches. Hybrid Model 2 achieved the best holdout performance, obtaining the lowest RMSE and the highest R2 value. Statistical testing confirmed its superiority over classical forecasting models. Residual diagnostics showed limited systematic bias, although extreme UV radiation peaks remained the principal source of forecasting uncertainty. These findings provide a reproducible artificial intelligence framework for short-term UV radiation forecasting and support intelligent early warning systems for public health protection, environmental monitoring, and climate resilience, contributing to Sustainable Development Goal 13 on Climate Action.

Algorithms doi: 10.3390/a19070521

Authors: Stefan Panić Milan Dubljanin Milan Savić Marko Smilić

This paper introduces an adaptive A-semilogarithmic gradient quantization framework aimed at reducing memory overhead and computational complexity during the training of deep neural networks. The approach employs a semilogarithmic companding function parameterized by a dynamically adjusted scaling factor A, which evolves in response to the statistical properties of gradients throughout the training process. Two distinct quantization strategies are proposed and evaluated: The switching piecewise A-quantizer, which adaptively toggles between low-bit uniform and high-bit semilogarithmic quantization according to an exponentially weighted moving-average (EMA) estimate of gradient variance; and the hybrid A-quantizer, which statically partitions the gradient domain, applying uniform quantization in low-magnitude regions and semilogarithmic companding in high-magnitude regions. The proposed methods are empirically evaluated on both multilayer perceptron (MLP) and convolutional neural network (CNN) architectures using tabular and image-classification benchmarks, including DCCC, CIFAR-10, CIFAR-100, and ImageNet. Quantitative results demonstrate that both models achieve comparable classification accuracy to full-precision (FP32) baselines while significantly reducing gradient reconstruction error. Notably, the hybrid A-quantizer consistently yields better validation accuracy, reduced RMSE, and improved convergence behavior relative to its switching counterpart. These findings underscore the effectiveness of hybrid semilogarithmic quantization as a robust and efficient solution for training deep models in resource-constrained or bandwidth-limited environments, with strong potential for scalable deployment across diverse hardware platforms.

Algorithms doi: 10.3390/a19070520

Authors: Abderrahim Sakouchi Habib Benbouhenni Nicu Bizon

The increasing demand for clean, reliable, and sustainable energy has intensified the need for advanced control strategies in modern wind energy conversion systems. Although conventional backstepping control (BC) offers strong stability and robustness, its performance may deteriorate under parameter uncertainties and dynamic operating conditions, leading to power fluctuations and reduced energy quality. To overcome these challenges, this study proposes an intelligent fuzzy fractional-order BC (FFOBC) strategy for multi-machine wind energy systems. By integrating fuzzy logic with fractional-order calculus into the classical BC framework, the proposed approach enhances adaptability, dynamic response, and robustness against system disturbances and nonlinearities. The controller is implemented at the machine-side inverter and validated in MATLAB/Simulink under varying wind and load conditions. Comparative results demonstrate that the proposed FFOBC significantly outperforms conventional sliding mode control in terms of overshoot reduction, steady-state accuracy, response smoothness, and total harmonic distortion minimization. Furthermore, the proposed strategy improves energy conversion efficiency, reduces mechanical and electrical stress, and ensures stable power injection into the grid. These findings highlight the potential of the proposed intelligent control framework to support sustainable, resilient, and high-quality wind energy integration in future smart power systems.

Algorithms doi: 10.3390/a19070519

Authors: Bingwei Cao Baoqing Yu Jiaxin Jiang Jiaqi Dong

Pressure fluctuations in the articulated steering system of wheel loaders can degrade steering smoothness, operational stability, and energy utilization efficiency. To address this issue, this study starts from the stroke difference and force-arm difference of steering cylinders induced by articulation motion, systematically reveals the structural mechanism responsible for pressure fluctuations in the steering mechanism, and proposes a suppression method based on hinge-point optimization. Specifically, a mathematical model of the articulated steering mechanism is established according to the analytical relationships between the stroke difference, force-arm difference, and articulation angle. The Dung Beetle Optimizer (DBO) is introduced to optimize and compare the hinge-point coordinates of the steering cylinders under different single-objective and multi-objective functions, thereby clarifying that the force-arm difference is the dominant factor affecting pressure fluctuations. Prototype modification and full-vehicle experiments are then conducted for validation. The results demonstrate that the hinge-point coordinates optimized with the force-arm difference as the objective function can significantly suppress steering pressure fluctuations. This study provides a theoretical basis and engineering reference for structural design, hinge-point layout optimization, and pressure-fluctuation suppression in articulated steering systems of wheel loaders.

Algorithms doi: 10.3390/a19070518

Authors: Fengyi Jin Yanli Liu

Structural optimization plays a crucial role in enhancing the performance of magnetic actuators. Traditional design approaches, such as parametric scanning, are limited by their reliance on empirical geometries. Meanwhile, widely used topology optimization techniques—including the Solid Isotropic Material with Penalization (SIMP) method and level-set methods—often encounter difficulties such as a large number of design variables, high computational cost, and unclear structural boundaries. To overcome these limitations, this paper proposes a novel gradient-free topology optimization method based on superellipses for designing magnetic actuator yokes. The proposed approach offers three key benefits: (1) It requires very few design variables, with each superellipse described by only seven parameters, thereby reducing the dimensionality of the design space and simplifying the optimization problem. (2) It yields clear and smooth structural boundaries without the need for post-processing. (3) It operates without gradient information, employing stochastic algorithms such as genetic algorithms that rely solely on objective function evaluations. A case study on yoke optimization demonstrates that our method achieves magnetic force output comparable to or better than the SIMP method, but with significantly fewer variables and a simpler implementation. This work provides an efficient and new tool for the conceptual design of magnetic actuators and related electromagnetic devices.

Algorithms doi: 10.3390/a19070517

Authors: Jingyu Hu Hongbo Bo Jun Hong Xiaowei Liu Weiru Liu

Graph Neural Networks (GNNs) often suffer from degree bias in node classification tasks, where prediction performance varies across nodes with different degrees. Several approaches, which adopt Graph Contrastive Learning (GCL), have been proposed to mitigate this bias. However, the limited number of positive pairs and the equal weighting of all positives and negatives in GCL still lead to low-degree nodes acquiring insufficient and noisy information. This paper proposes the Hardness Adaptive Reweighted (HAR) contrastive loss to mitigate degree bias. It adds more positive pairs by leveraging node labels and adaptively weights positive and negative pairs based on their learning hardness. In addition, we develop an experimental framework named SHARP extending HAR to a broader range of scenarios. Both our theoretical analysis and experiments validate the effectiveness of SHARP. Across four datasets, SHARP outperforms baselines in 14 of 16 settings, improving global accuracy on Cora by 3.6% and 4.2% under the GCN and GAT backbones, while at the degree level it raises accuracy for the lowest-degree nodes by over 10%, confirming that the gains are targeted at the low-degree nodes most affected by degree bias.

Algorithms doi: 10.3390/a19070516

Authors: Balázs András Tolnai Zheng Grace Ma Bo Nørregaard Jørgensen

Non-intrusive load monitoring (NILM) estimates component-level energy use from aggregate measurements, but low-frequency data limit appliance signatures and make overlapping or weakly observed loads difficult to separate. This paper proposes a relation-aware multi-driver pipeline for interpretable low-frequency load attribution under partial observability. The method does not require supervised component labels or predefined appliance models. It combines semantic feature typing, heterogeneous relation discovery, feature-family construction, mechanism-aware evidence modeling, conservative allocation, event-background separation, and role-based attribution. Only evidence-supported load is assigned to feature families, while unsupported variation is retained as unexplained demand or residual load. The method is evaluated in a simulated EV-focused building case and through measured-building validation on nine ADRENALIN buildings. In the EV case, the selected EV-aligned family achieved a correlation of 0.990 and an NMAE of 0.100 against the withheld EV reference, while heat-pump and base-load recovery was weaker, with NMAE values of 0.565 and 0.895. In the ADRENALIN validation, temperature-associated families achieved median NMAE values of 0.594 using the restricted feature set and 0.576 using the full feature set. Additional comparison, ablation, sensitivity, diagnostic, and runtime analyses show that the pipeline is most effective for dominant event-driven loads, remains limited for smoother or masked lower-magnitude components, and treats unexplained variation explicitly. The results demonstrate a practical framework for interpretable driver-based load attribution when component labels are unavailable or incomplete.

Algorithms doi: 10.3390/a19070515

Authors: Zina Abohaia Patrick Mukala

The rapid growth of digital data has created an urgent need for innovative storage solutions. Transactional data, with its diverse formats and row-based structure, is a significant contributor to this challenge. Existing storage methods struggle to compress heterogeneous structured data efficiently due to limited redundancy exploitation across mixed data types. This study introduces Pixelated, a novel lossless data transformation and storage framework that converts structured transactional data into pixel representations stored in Portable Network Graphics (PNG) format. Pixelated introduces a new data representation strategy that enables the existing DEFLATE compression mechanism within PNG to exploit patterns and redundancy in heterogeneous transactional datasets more effectively. Designed for datasets containing numerical, categorical, and datetime values, Pixelated achieves average compression rates exceeding 90%. The framework is evaluated across benchmark and real-world datasets, demonstrating competitive performance against ZIP, Apache Parquet, and Python Pickle. Detailed methodology, redundancy characteristics, limitations, and performance results are presented.

Algorithms doi: 10.3390/a19070514

Authors: Luís Henrique Travassos Edite Ravella Jorge Bernardino Francisco B. Pereira

Human Activity Recognition (HAR) is the automated process of identifying human actions using sensor data or video, which is widely used in healthcare, smart environments, and surveillance. Although HAR based on computer vision has advanced rapidly, existing reviews do not adequately address the recent shift toward hybrid deep-learning architectures or provide a structured comparison of the trade-offs relevant to real-world deployment. This literature review addresses that gap through a PRISMA-guided analysis of articles published between 2021 and 2025 and retrieved from four major databases. The review develops a reproducible taxonomy of nine architectural families and applies a multidimensional evaluation framework covering classification accuracy, computational efficiency for edge deployment, environmental generalization, and fine-grained activity recognition. The findings show that hybrid architectures are the dominant design strategy, while attention-based and graph-based models play important specialized roles depending on temporal complexity, privacy requirements, and deployment constraints, with the literature concentrated mainly in healthcare and security applications.

Algorithms doi: 10.3390/a19070513

Authors: Lan Jiang Yesen Wang Yingchun Zhang Zijun Xiang Xinhong Han

To balance the requirements of structural safety and economic efficiency in transmission tower design, this study proposes a structural optimization method based on an Improved Sparrow Search Algorithm (ISSA). Specifically, the methodology integrates Circle chaotic mapping to initialize the sparrow population distribution, followed by the introduction of a firefly perturbation strategy. This strategy updates the sparrow positions derived from the basic algorithm, thereby enhancing the algorithm’s ability to escape local optima. Based on these enhancements, a mathematical model for structural optimization is established with the primary objective of minimizing the total weight of the tower. The penalty function method is employed to handle constraints, while the finite element method is utilized for internal force analysis. To verify the effectiveness of the proposed approach, a typical cat-head transmission tower is selected as a case study. The optimization results obtained from the ISSA are compared with those produced by the Daoheng software, demonstrating a weight reduction of 3.79%. These findings indicate that the ISSA effectively improves the optimization performance and provides an efficient solution for the preliminary sizing design of transmission towers.

Algorithms doi: 10.3390/a19070512

Authors: Xinhao Lu Jiuxin Cao Hao Chen

High-penetration integration of Renewable Energy Sources (RESs) and Battery Energy Storage Systems (BESSs) has transformed Distribution Network Reconfiguration (DNR) from a static topology optimization task into a complex combinatorial problem with strong coupling between discrete switch decisions and dynamic power flow constraints. This evolution requires that optimization algorithms provide two capabilities: native adaptability to discrete decision variables without approximation, and real-time responsiveness to dynamic operating conditions. Ant Colony Optimization (ACO), as the most widely applied discrete-native Swarm Intelligence (SI) algorithm, faces three critical bottlenecks in DNR due to its Traveling Salesman Problem (TSP)-oriented design: framework incompatibility, ambiguous heuristic formulation, and ineffective pheromone strategies. To address these limitations, this study proposes a Graph-Theoretic Ant Colony Optimization (GTACO) algorithm. Multi-scenario experiments on IEEE 33-bus and PG&E 69-bus systems demonstrate that GTACO outperforms state-of-the-art algorithms in core metrics including loss reduction rate, voltage stability, convergence efficiency, and economic–environmental benefits. This research overcomes the TSP-centric limitations of conventional ACO, establishes a methodological foundation for extending the ACO framework to complex non-TSP discrete optimization tasks, and provides a practical solution for dynamic DNR under high-penetration RES and BESS integration.

Algorithms doi: 10.3390/a19070511

Authors: Ziye Wang Jianyu Xiao

The creation of Basic Probability Assignment (BPA) still represents a basic problem in the Dempster-Shafer (D-S) theory of evidence especially when it comes to representing continuous uncertainty and class ambiguity. In order to overcome this problem, this paper suggests a BPA construction model depending on Gaussian overlap. The main principle behind the approach is the creation of focal elements based on the overlaps between conditional probability distributions of classes, allowing characterisation of uncertainty in a data driven manner. Namely, attribute level evidence is represented by Gaussian distributions, and singleton and composite focal elements are composite focal elements are generated through Gaussian product responses and normalized to obtain BPAs. Composite focal elements are further projected into singleton-level decision scores through proportional belief and plausibility transformations for decision-making and attribute-weight calculation. Moreover, to dynamically modify the role played by different attributes, a Kullback-Leibler (KL) divergence-based weighting scheme is used. These parts combine to form a full pipeline of continuous evidence modeling to BPA generation as proposed by the given method. The experimental results show that the proposed method achieves 98.00 ± 2.67% accuracy on the Iris dataset, 97.21 ± 1.76% accuracy on the Wine dataset, and 90.86 ± 1.20% accuracy on the Breast Cancer Wisconsin dataset. Compared with existing BPA generation methods, the proposed method obtains the best performance on the Iris and Wine datasets. Compared with classical machine learning models, the method also achieves the highest accuracy on the Iris dataset and remains competitive on the Wine and Breast Cancer Wisconsin datasets.

Algorithms doi: 10.3390/a19070510

Authors: Nikola Komatina Dragan Marinković Vladimir Simić Nikola Banduka Aleksandar Nešović

This study developed a hybrid fuzzy decision-making algorithm based on the Fuzzy Best-Worst Method (FBWM) and the Fuzzy Square-Root-based Evaluation Method (FSREM). Despite the widespread application of the 8D methodology in engineering practice, the importance of its disciplines has not been sufficiently investigated; therefore, the aim of this study is to determine their significance and priority. The proposed fuzzy algorithm was applied to three companies operating within the automotive supply chain. FBWM was used to determine the criteria weights, while FSREM was applied to rank the 8D disciplines. Sensitivity analysis showed that the expert teams from the three considered companies perceived the problem in a very similar manner. The results of applying the proposed algorithm in all three companies showed that the discipline Identify and Verify Root Cause (D4) has the greatest influence on problem-solving effectiveness. In two of the three companies, Prevent Recurrence (D7) was ranked as the second most influential discipline, while in one company Define Permanent Corrective Actions (D5) was identified as the second most influential discipline. It can be concluded that the results demonstrated a high degree of consistency, while minor ranking deviations can be attributed to different quality management system approaches within each company.

Algorithms doi: 10.3390/a19070509

Authors: Stamatis Chatzistamatis George E. Tsekouras Anastasios Rigos Alvaro Garcia-Recuero Eleni Valari Andreas Siafakas Konstantinos Kotis

Recidivism prediction is increasingly embedded in criminal justice decision-making, yet most deployed systems remain opaque and have been shown to exhibit discriminatory behavior against certain demographic groups. This paper presents a fairness-aware interpretable framework for recidivism prediction applied to three real-world datasets from Bulgaria, Greece, and Portugal. The classification core relies on a 1-Dimensional Convolutional Neural Network (1D-CNN), trained by a custom objective function that embeds the Equalized Odds fairness criterion as an L1-regularized penalty reflecting on gender-based disparities in false positive and false negative rates. Model-level interpretability is provided through Kernel SHAP, which decomposes individual predictions into additive feature attributions grounded in cooperative game theory. Experiments across prediction tasks, each evaluated over randomized runs, demonstrate that the baseline model exhibits statistically significant bias against the female group in all datasets. The fairness-constrained model substantially reduces these disparities across all tasks at a moderate and expected cost to classification accuracy. Kernel SHAP analysis reveals the relative contribution of static and dynamic offenders’ attributes to individual risk scores, supporting auditability and contestability. The proposed framework advances the integration of predictive performance, algorithmic fairness, and structural interpretability in criminal justice analytics.

Algorithms doi: 10.3390/a19070508

Authors: Xiaoyuan Fu Li Huang Weiwei Du Yuqi Jin

Community multi-building energy management is a key approach for reducing carbon emissions from the building sector and alleviating peak grid pressure. However, load coupling among buildings and coordinated energy-storage operation make control-policy design highly challenging. To address the limitation of the standard multi-agent proximal policy optimization (MAPPO) algorithm, in which the centralized critic simply concatenates building observations and therefore struggles to model inter-building interactions, this paper proposes an improved MAPPO algorithm with a multi-head-attention-enhanced centralized critic, referred to as a multi-head-attention MAPPO (MHA-MAPPO). Without changing the decentralized execution framework, the proposed method improves the critic network in three aspects. First, a dual-branch gated embedding module is designed to adaptively fuse local building observations and global interaction information. Second, an interaction-attention path is constructed to explicitly capture pairwise dependencies among buildings through multi-head attention. Third, a context-attention path is introduced to extract high-level community-level global features by means of learnable query vectors. These improvements enable the critic to estimate the joint-state value more accurately and provide more reliable advantage estimates for all agents. Experiments in the CityLearn environment show that, compared with the original MAPPO, MHA-MAPPO improves the mean evaluation reward by approximately 19.2%, reduces the reward standard deviation by one order of magnitude, and decreases peak net load and total net load by approximately 15.4% and 35.5%, respectively. The results verify the effectiveness of multi-head attention for coordinated multi-building scheduling. The proposed method provides a useful reference for improving multi-agent reinforcement learning algorithms in community energy management.

Algorithms doi: 10.3390/a19070507

Authors: Yuan Xu Enyong Xu Yingnan Gao Zhenzhen Jin

Rotating machinery is a key component in modern industry, and its operating condition directly affects equipment safety and production reliability. However, discrepancies among different machines cause source–target distribution shifts, while fault annotation for target machines is costly, limiting the performance of deep learning-based diagnosis under cross-machine scenarios with limited labels. To address these issues, this paper proposes a multi-scale time–frequency semantic consistency model based on self-supervised transferable representation learning, termed MTFSC. First, augmented waveform views and multi-scale frequency-domain views are constructed from unlabeled source-domain vibration signals for self-supervised pre-training without source labels. Then, a time-domain impulse-aware feature extractor and a time–frequency decoupled spectral feature extractor are designed to enhance local impulsive responses and emphasize fault-sensitive time–frequency patterns. Furthermore, a semantic-aware soft contrastive loss is developed to mine potential semantic neighbors from multi-scale frequency-domain structural similarity, reducing false-negative effects in conventional hard-label contrastive learning. Finally, the pre-trained time-domain extractor is transferred to the target machine and fine-tuned with limited labeled samples. Experimental results show that MTFSC outperforms comparison methods under different labeled sample ratios and achieves an average accuracy of 97.5% across four cross-machine diagnostic tasks.

Algorithms doi: 10.3390/a19070506

Authors: Ahmed Abdallah Abaker Khalid Aldriwish Ibrahim Rizqallah Alzahrani Daifallah Zaid Alotaibe

The health system is now more complex, uncertain, interdependent, and dynamically interconnected than ever, making traditional planning decisions based on static, reductionist models increasingly impracticable. Systems analytics approaches such as system dynamics, agent-based modeling, and network analysis are often deployed in isolation and fail to capture cross-level interactions and emergent system behavior. This study proposes an integrated multi-layer systems analytics framework that combines these analytical paradigms within a unified architecture to support adaptive health systems planning under uncertainty. The proposed framework introduces an Adaptive Policy Intelligence Layer (APIL), which enables continuous feedback-driven policy adaptation through dynamic monitoring, scenario evaluation, and real-time adjustment mechanisms. The model is evaluated under multiple simulation scenarios, including baseline conditions, demand shocks, resource constraints, and digital transformation environments. The findings provide strong quantitative and analytical evidence of improved system performance and resilience. More specifically, the digital transformation scenario achieved the lowest mean system pressure (0.128) and the highest resilience index (0.887), while the demand shock scenario produced the highest peak system pressure (0.306). The results demonstrate enhanced system resilience, more efficient resource deployment, and superior policy responsiveness compared with traditional single-method approaches. The originality of this study lies in integrating multi-level systems analytics with adaptive policy intelligence into a unified, feedback-driven decision-support framework for resilient health systems governance. The study contributes to systems analytics literature by advancing a synergistic and adaptive modeling paradigm capable of supporting policymakers in navigating complex and unstable healthcare environments.

Algorithms doi: 10.3390/a19070505

Authors: Beza Negash Getu Nuhamin Kifle Semu

Accurate waveform classification in noisy environments is an important task in modern communications, radar signal analysis, biomedical signal interpretation, industrial monitoring and other signal processing systems. This paper investigates the performance of three neural network architectures: Multilayer Perceptron (MLP), one-dimensional Convolutional Neural Network (1D-CNN), and Long Short-Term Memory (LSTM) for multiclass waveform classification in the presence of Additive White Gaussian Noise (AWGN). A time series dataset consisting of multiple waveform classes is generated and corrupted with AWGN across a wide range of signal-to-noise ratio (SNR) levels to simulate noisy signal distortion conditions. The three models are trained and evaluated under identical conditions to ensure a fair comparison. Their classification performance is evaluated in terms of accuracy, Confusion Matrix (CM), Receiver Operating Characteristic (ROC) curve and the Area Under the ROC curve (AUC) across varying SNR values. Simulation results demonstrate that the 1D-CNN effectively captures local temporal patterns and achieves superior robustness in classification at moderate and high SNR levels. The LSTM model demonstrates the ability to capture temporal dependencies in sequential waveform data but exhibits sensitivity to waveform variations due to amplitude, phase and frequency changes and noise at lower SNR values. The MLP, although computationally simpler, shows comparatively limited performance in low-SNR conditions due to its lack of temporal feature extraction capability. For the case of multiclass deterministic waveforms, the accuracy of classification for the 1D-CNN and LSTM is nearly 100% at SNR = 5 dB showing their robustness in classification, whereas the accuracy of MLP is approximately 70% that shows poor classification in noisy conditions. When there is random amplitude, frequency and phase variations in the waveforms, the accuracy of the 1D-CNN and MLP increases with SNR, and 1D-CNN superior to MLP. However, the LSTM accuracy fails to improve with SNR, resulting in poor classification performance in such a scenario. The results provide an insight into the suitability of different neural architectures for waveform classification tasks in noisy communication or other time series applications and highlight the advantages of convolutional feature extraction for robust signal recognition.

Algorithms doi: 10.3390/a19070504

Authors: Paolo Pagliuca

Evolutionary Strategies (ESs) are optimization metaheuristics largely adopted in Evolutionary Computation (EC). Since their introduction in the early 70s, researchers in the field have attempted to improve the efficacy of these algorithms. The most advanced ESs, such as the Covariance Matrix Adaptation Evolutionary Strategy (CMA-ES) and Exponential Natural Evolution Strategies (xNESs), make use of covariance matrices storing relationships between parameters to be optimized, which enable the algorithms to fasten the search in the solution spaces. However, the computational cost of calculating covariance matrices linearly scales with the number of parameters. Recently, the OpenAI Evolutionary Strategy (OpenAI-ES) emerged as an effective ES in different domains, thanks to the parameter information stored in two momentum vectors. Furthermore, OpenAI-ES gains an advantage from the usage of symmetric sampling and weight decay techniques. In this work, I delve into the application of symmetric sampling and weight decay on CMA-ES, xNES and Separable Natural Evolution Strategies (sNESs), with the aim to improve their performance in domains in which they get stuck in local minima outcomes. Specifically, I propose three novel variants for each ES and verify their efficacy with respect to the PyBullet halfcheetah and hopper robot locomotion problems, and two collective tasks (i.e., swarm aggregation and swarm foraging). The findings reveal that symmetric sampling produces performance enhancements in all the domains, whereas the effect of weight decay varies across the considered problems. Furthermore, symmetric sampling allows ESs to keep parameter size limited, which is paramount in these scenarios. This research identifies techniques enhancing the success of modern ESs, proposes several ES variants, and discusses the relationship between algorithmic performance and task properties.

Algorithms doi: 10.3390/a19070503

Authors: Shupeng Fang Senlin Chen Yiyun Zhao Sijie Yao

Trajectory optimization during the terminal area energy management (TAEM) phase is pivotal for achieving accurate runway alignment and enhancing landing safety in autonomous aircraft operations. In the presence of initial state uncertainties in TAEM phase, conventional pseudo-spectral methods still suffer from robustness limitations and exhibit a strong dependence on the quality of the initial guess. Therefore, this paper proposes the composite African vulture optimization algorithm (CAVOA), a meta-heuristic framework designed to automate trajectory optimization. An in-depth examination of the heading alignment cone (HAC) trajectory model enables effective heading adjustments prior to landing, augmented by a tailored dynamic pressure profile to ensure safe touchdown velocities. By incorporating dynamic opposition learning, intelligent boundary processing, and composite exploration, CAVOA enhances global search efficiency. These enhancements are substantiated through comparisons with benchmark function optimization, Wilcoxon rank sum tests, and convergence analysis. Numerical simulations validate that CAVOA reliably directs autonomous aircraft to predefined touchdown states, demonstrating superior performance in complex aerial environments.

Algorithms doi: 10.3390/a19070502

Authors: Jaime Gallardo-Alvarado Alvaro Sanchez-Rodriguez Horacio Orozco-Mendoza Ramon Rodriguez-Castro Luis A. Alcaraz-Caracheo

This work presents the kinematic analysis of a redundant planar parallel manipulator within the framework of screw theory. The main contribution of this work is the introduction and kinematic modeling of a novel redundant planar parallel manipulator topology composed exclusively of revolute joints. The proposed architecture is motivated by the search for structurally simple mechanisms with favorable analytical properties for screw-theoretic formulation and potential applications in robotic systems requiring compact and efficient planar motion. For completeness, the displacement analysis is included. Thanks to the simple topology of the otherwise complex mechanism, the inverse–forward displacement problem is resolved through straightforward quadratic equations. The velocity input–output relationship is derived without reliance on passive joint rate velocities, and the acceleration input–output equation is obtained independently of passive joint rate accelerations. These simplifications are achieved by exploiting reciprocal line properties. Numerical examples are provided to illustrate the robustness and effectiveness of the proposed kinematic analysis method across the main topics addressed in this contribution.

Algorithms doi: 10.3390/a19060501

Authors: Yanyang Hou Shufeng Xiong Yang Li

Chinese cloze question generation for educational assessments requires identifying gap phrases that accurately reflect key knowledge points, posing significant challenges to automated systems. We observe that the syntactic boundaries revealed by part-of-speech (POS) tags closely align with the semantic boundaries of target gap phrases. Motivated by this observation, we propose a multi-task learning framework in which gap phrase identification serves as the primary task and POS tagging as a complementary auxiliary task. The two tasks share a common BERT-BiLSTM encoder, enabling mutual reinforcement of both syntactic and semantic representations through joint training. To further capture the interaction between label semantics and contextual word representations, we introduce a label-attention mechanism that models dependencies between the global word sequence and candidate label embeddings. Additionally, we construct a refined POS tag subset by excluding categories whose boundaries show no alignment with gap phrase boundaries, thereby strengthening the correspondence between the two tasks. Evaluated on a real-world dataset of 20.5K questions spanning five academic disciplines, our method achieves an F1 score of 65.85%, with a Recall of 67.79%, representing improvements of 2.12% and 4.35% over the prior state-of-the-art, respectively. These results demonstrate that exploiting the alignment between syntactic and semantic structures through joint learning is effective for generating educationally meaningful fill-in-the-blank questions.

Algorithms doi: 10.3390/a19060500

Authors: Shuwen Liang Zhi Qiao Tianyu Bai Ying He Dong’er Chen Song Fu

LLM privacy risks arise across different lifecycle stages and architectural boundaries, and existing protection mechanisms provide only partial coverage. This paper analyzes the main families of privacy-preserving approaches for LLM systems through a two-axis structure that crosses lifecycle stages with system architecture layers. Some safeguards are operationally mature; others, such as confidential computing, have moved into production practice; stronger cryptographic methods, while most promising in principle, remain research-heavy in practice. No single mechanism provides complete end-to-end protection: different methods protect different assets, operate at different lifecycle stages, span distinct system layers, and carry distinct trust, performance, and deployment trade-offs. Practical LLM privacy is therefore a problem of layered system design rather than the search for a universal primitive, and hybrid architectures are emerging as the most realistic deployable pattern. Building on this analysis, we propose a six-dimensional evaluation framework for privacy in hybrid LLM deployments (a 0–5 ordinal scoring rubric designed for reproducible application, with explicit anchor language and per-score evidence requirements) and apply it to five representative confidential AI deployments, deriving the scores in full for two of them. The framework feeds a three-tier gap-closure roadmap and design principles for architecture-time use, connecting what privacy technologies promise, what they actually protect, and what is realistically deployable in modern LLM systems.

Algorithms doi: 10.3390/a19060499

Authors: Chun Xiao Xiaoqing Han Tingjun Li

With the large-scale integration of renewable energy and the deepening of electricity market reform, uncertainty in power system operation has increased significantly. This creates new challenges for multiple generation companies when they work together to develop generation schedules that balance economic efficiency and low-carbon goals. Most existing studies assume fixed loads and ignore the active regulation capability of the demand side under price signals and incentive signals. To address this gap, this paper proposes a low-carbon generation schedule optimization method for multiple generation companies. The method considers heterogeneous flexible loads. First, the paper decomposes flexible load adjustability into two components: price elasticity-based load shifting and incentive-based adjustable capacity. Using the price elasticity matrix method, the market clearing price serves as a known input. The load shifting amount under price elasticity regulation is pre-calculated for each park and treated as an exogenous parameter in the generation schedule model. This allows generation companies to directly use demand-side flexibility information during the planning stage. Second, the paper uses the proportion of residential and industrial loads as a core parameter. It characterizes the heterogeneity of four parks along two dimensions: elasticity coefficients and upper limits of adjustable capacity. Parks with a higher proportion of industrial loads have stronger flexible regulation capability. This result is consistent with real physical characteristics. It also provides a quantitative basis for generation companies to utilize flexible resources differently across parks and optimize their output arrangements. Finally, the paper uses the upward and downward adjustable capacity of each park as decision variables. It builds a multi-generator low-carbon generation schedule optimization model with heterogeneous flexible loads. Generator output constraints, power balance constraints, flexible load adjustable capacity constraints, and carbon quota constraints are all integrated into a single-level mixed-integer linear programming framework. This framework can be solved efficiently using commercial solvers. It helps generation companies develop optimal generation schedules that balance economic efficiency and low-carbon targets. Case study results show that combining price elasticity regulation with incentive-based adjustable capacity can effectively improve both the economic performance and low-carbon performance of generation schedules.

Algorithms doi: 10.3390/a19060498

Authors: Qinglei Zhang Yao Wang Ying Zhou

To address the insufficient accuracy of short-term load forecasting caused by the strong randomness of distributed energy output, variable electricity consumption patterns, and complex meteorological factors, this study proposes a load forecasting method that integrates K-means scene clustering and a Transformer–BiGRU–Attention (CTBA) hybrid deep learning architecture. Different from conventional Transformer–BiGRU hybrid forecasters that train a single global predictor across all operating conditions, the proposed CTBA framework first partitions daily load curves into representative scenes and then routes each sample to a scene-specific Transformer–BiGRU–Attention predictor, thereby reducing distributional heterogeneity before temporal modeling. First, the K-means algorithm is used to perform scene clustering on historical daily load curves, and the optimal number of clusters is selected according to the silhouette coefficient and downstream prediction performance. Subsequently, the CTBA model is trained separately for each clustering subset. The Transformer encoder captures the long-range global dependencies of load sequences through the self-attention mechanism, the BiGRU module extracts local bidirectional temporal fluctuation features, and the Attention mechanism further focuses on key time nodes such as morning and evening peaks while fusing multi-source data including historical load, day-ahead electricity price, and multi-dimensional meteorological factors. Experimental results based on the German ENTSO-E power dataset show that the coefficient of determination R2 of the proposed model reaches 0.9893, with MAE, RMSE, and MAPE as low as 0.0141, 0.0187, and 3.92%, respectively, which are significantly improved compared to benchmark models such as SVR, LSTM, CNN, and TCN-BiGRU. Ablation experiments further demonstrate that removing the clustering, Transformer, BiGRU, or attention layer will degrade performance, thus verifying the effectiveness and superiority of the method in short-term load forecasting and providing an accurate solution for the short-term load forecasting of power systems.

Algorithms doi: 10.3390/a19060497

Authors: Mlungisi Ntombela

Electric Vehicle Charging Stations (EVCSs) have become increasingly important due to the growing penetration of electric vehicles (EVs) and renewable-based power generation. However, challenges such as fluctuating renewable energy availability, increasing charging demand, power losses, operational cost, and charging delays continue to affect overall grid performance and stability. To address these issues, this study proposes a hybrid Goat Optimization Algorithm–Hierarchical Multi-Granularity Interaction Graph Convolutional Network (GOA–HMGIGCN) framework for intelligent smart grid energy management and EV charging coordination. The proposed framework combines the Goat Optimization Algorithm (GOA) for optimal EVCS placement and charging scheduling with the Hierarchical Multi-Granularity Interaction Graph Convolutional Network (HMGIGCN) for forecasting renewable generation, charging demand, and load variations. The framework was implemented and evaluated in MATLAB/Simulink R2024a using the IEEE 14-bus smart grid test system under varying operating conditions. Simulation results demonstrated that the proposed framework achieved superior performance compared with the Coot Optimization Algorithm–Fractional Backpropagation Physics-Informed Neural Network (COA-FBPINN), Dingo Optimization Algorithm–Convolutional Hypergraph Graph Neural Network (DOA-CHGNN), Self-Feedback Feedforward Artificial Neural Network (SFFANN), Deep Neural Network (DNN), and Golden Jackal Optimization–Attention-Based Probabilistic Convolutional Neural Network (GJO-APCNN) techniques by attaining the lowest operational cost of USD 1561, the highest efficiency of 99.2%, the minimum power loss of 10.6 kW, and the shortest charging time of 32 min. In addition, the proposed framework and overall grid reliability, confirming its effectiveness for intelligent renewable-integrated smart grid applications.

Algorithms doi: 10.3390/a19060496

Authors: Luis Fernando Bustos-Marquez Steven Hegedus

Accurate plant-level photovoltaic (PV) power estimation is important for performance monitoring, model benchmarking, and grid-integration studies. In bifacial horizontal single-axis tracking (HSAT) systems, this task is complicated by the coupled effects of front-side irradiance, rear-side irradiance, tracker position, and module temperature. This study proposes an algorithmic framework for same-time-step AC power estimation in a bifacial HSAT PV plant using field measurements of irradiance, tracker angle, module temperature, and inverter active power. The framework is not intended as an operational forecasting model because future irradiance and weather conditions are not predicted; instead, it evaluates how compact physics-based structure, interpretable nonlinear learning, and ensemble learning estimate measured AC power under nominal operating conditions. An empirical rear-to-front irradiance relationship was derived using solar-elevation bins and incorporated into a compact physics-based benchmark. This benchmark was compared with an additive Explainable Boosting Machine (EBM) and a Random Forest (RF) on a common test subset of 3916 observations. The physics-based model achieved an RMSE of 19.6 kW, an R2 of 0.72, and an NRMSE of 0.38. The EBM improved these values to 17.09 kW, 0.786, and 0.334, respectively, while the RF achieved 15.96 kW, 0.814, and 0.312. Chronological validation showed weaker and more variable performance than randomized validation, indicating that temporal generalization remains challenging. Overall, the results support the use of interpretable PV-domain-guided learning as a transparent intermediate approach between compact physics-based modeling and more flexible ensemble regression.

Algorithms doi: 10.3390/a19060495

Authors: Shanshan Peng Dandan Wang

In the fabric warehouse, order picking operations face high labor intensity and rising operational costs, requiring urgent optimization. This study investigates the order scheduling and task assignment problem within an elastic staffing framework, where temporary labor recruitment and real-time task allocation need to be adjusted dynamically in response to fluctuations in order volumes. Nevertheless, conventional approaches often suffer from severe computational bottlenecks under such highly dynamic conditions, and struggle to maintain optimal solutions when demand undergoes large and frequent fluctuations. To address these challenges, this study proposes a Graph Transformer Policy Network with Proximal Policy Optimization (GTP-PPO), which combines graph structure features with a global attention mechanism. First, the return picking strategy and the S-shaped picking strategy are compared and analyzed in the fabric warehouse scenario. The results reveal that the return strategy is more suitable for the studied warehouse layout. Subsequently, a mixed-integer programming (MIP) model and a GTP-PPO model are established for optimizing order dispatching and scheduling. Finally, an empirical analysis is carried out based on the peak order day of the year in the fabric warehouse. The results demonstrate that the proposed GTP-PPO model not only achieves near-global optimal solutions (gap < 4%) comparable to the MIP model, but also exhibits robust real-time decision-making capabilities under dynamically increasing order volumes and unexpected disruptions. Compared to the MIP model, the GTP-PPO approach reduces unskilled labor hours by 84.80% and decreases operational volatility by 27.60%, with only a 3.52% increase in operational costs.

Algorithms doi: 10.3390/a19060494

Authors: Jiuxia Guo Zihao Ren Yaqian Du Jingyang Huang Pengcheng Dan

Aircraft taxiing speed control along predefined airport surface routes is a constrained single-aircraft longitudinal control problem involving heterogeneous route geometry, action smoothness, and terminal velocity feasibility. Existing learning-based taxiing controllers commonly use a unified policy for the whole route, which may be insufficient for handling straight-segment propulsion, curved-segment speed regulation, and action discontinuities near straight–curve transitions. This paper proposes SegCoord-Taxi, a geometry-aware segmented deep reinforcement learning framework for taxiing speed control. The route is decomposed into straight segments, curved segments, and transition boundary zones. A Straight-Segment Policy (SSP) and a Curved-Segment Policy (CSP) generate geometry-dependent base acceleration commands, a Switch Residual Adapter (SRA) provides local residual correction near transition regions, and a Route-Level Feasibility Projection (RFP) maps the coordinated action into an executable acceleration satisfying route-level feasibility constraints. Experiments on departure taxiing routes at Chengdu Tianfu International Airport (ZUTF) included baseline comparison, ablation analysis, projection diagnostics, sensitivity analysis, and a trajectory-level case study. On the evaluated ZUTF case-study routes, SegCoord-Taxi achieves the lowest final velocity on the test set, 0.336±0.017 m/s, compared with 0.732±0.061 m/s for the unified Proximal Policy Optimization (PPO) controller and 0.586 m/s for the curvature-aware constrained optimizer. The complete framework also reduces switch action jump from 1.022±0.017 m/s2 to 0.429±0.004 m/s2 in the ablation study. These results indicate improved terminal feasibility and transition-region smoothness in the evaluated single-airport case-study setting under an explicit efficiency–smoothness–feasibility trade-off. Future work will extend the framework to multi-aircraft and multi-airport settings under operational uncertainty.

Algorithms doi: 10.3390/a19060493

Authors: Rasha Hasan Rafe Alasem Ahmed Mahmoud Yazeed Alsarhan Mahmud Mansour

Oil and gas pipelines are critical infrastructures that require continuous and reliable monitoring to detect leaks, pressure anomalies, corrosion, and unauthorized activities. Wireless sensor networks (WSNs) have emerged as an effective solution for large-scale pipeline monitoring due to their low deployment cost and real-time sensing capabilities. However, the resource-constrained nature of sensor nodes and the open wireless communication environment expose pipeline monitoring systems to various routing attacks, for example, blackhole, sinkhole, selective forwarding, and false data injection attacks, while simultaneously demanding strict energy efficiency to prolong network lifetime. In this paper, we propose SEER-PM (Secure and Energy-Efficient Routing for Pipeline Monitoring): a novel protocol that integrates an Artificial neural network (ANN)-based trust mechanism with energy-aware routing metrics. SEER-PM dynamically evaluates node trustworthiness based on packet forwarding behavior, residual energy, and signal consistency. By training the ANN on historical behavioral data, the system accurately detects malicious nodes with high precision. Simulation results demonstrate that SEER-PM outperforms existing secure routing protocols (Sec-AODV and T-LEACH) in terms of packet delivery ratio (PDR) by 14%, detection rate by 9.5%, and network lifetime by 12% under heavy attack scenarios. The proposed protocol enhances the reliability, security, and sustainability of pipeline monitoring WSNs operating in harsh and remote environments.

Algorithms doi: 10.3390/a19060492

Authors: Maikel Leon

Artificial intelligence (AI) has reshaped higher education over six decades, evolving from drill-and-practice programs to adaptive cognitive tutors and, most recently, transformer-based generative models. This article presents a critical review of publisher-built generative AI assistants, adopting an explicitly socio-technical perspective that combines a technological lens with a pedagogical one. It makes three contributions. First, it synthesizes the technical and algorithmic evolution of educational AI, from rule-based and expert systems through knowledge tracing and learning analytics to large language models and retrieval-augmented generation, and organizes these mechanisms into a taxonomy. Second, it introduces a reproducible evaluation framework that couples the TRIAD rubric (Trust, Relevance, Impact, Adoption, and Design) with a Jobs-to-Be-Done (JTBD) lens, complete with anchored scoring criteria, an evidence-and-confidence grading scheme, and reported inter-rater reliability. Third, it applies the framework to eleven assistants released by U.S. publishers, distinguishing peer-reviewed evidence from institutional reports and commercial claims. The analysis reflects a mid-2025 snapshot and is presented as a reusable template rather than a static ranking. Findings reveal substantial variation in privacy safeguards, curricular alignment, documented impact, adoption, and usability. The review identifies application scenarios and recommendations for researchers and institutional leaders seeking to guide the responsible integration of AI in higher education.

Algorithms doi: 10.3390/a19060491

Authors: Arnold Kamis

This review examines the evolving intersections of data analytics, machine learning, and artificial intelligence—terms that have been frequently conflated since 2016 during a period of increased hype and investment. Following recent reviews across areas such as open innovation, supply chain deep learning, strategic alliances, natural language processing, and big data streaming, we focus on the emerging field of Responsible Artificial Intelligence (AI). We apply descriptive analysis to identify trends, patterns, and gaps in the research through a review of academic literature from 2020 to 2025. Analysis reveals five distinct clusters of Responsible AI papers using five dimensions: fairness, cross-validity, transparency, accuracy–interpretability tradeoff, and drift detection. This review discusses patterns across the artificial intelligence literature and identifies future research opportunities with an emphasis on Responsible AI.

Algorithms doi: 10.3390/a19060490

Authors: Aksultan Mukhanbet Paulo Trigo Beimbet Daribayev Darkhan Akhmed-Zaki

Quantum convolutional neural networks (QCNNs) are emerging as promising models in quantum machine learning, particularly for image classification and computer vision tasks. Recent developments include hybrid classical–quantum architectures, advanced quantum encoding methods, and novel circuit designs that improve data processing on Noisy Intermediate-Scale Quantum (NISQ) devices. However, practical implementation remains challenging due to circuit complexity, gate count, qubit connectivity, and hardware noise, which limit scalability and performance. Consequently, quantum circuit optimization has become essential for reducing resource requirements and improving classification accuracy. This study presents a systematic literature review of 40 research papers published between 2014 and 2025. The review covers QCNNs together with closely related quantum neural network (QNN) models and quantum circuit optimization studies, since circuit-optimization techniques are frequently developed for QNNs more broadly rather than for QCNN architectures in isolation. Within this scope, it examines network architectures, encoding strategies, application domains, and optimization techniques, with particular attention to heuristic and metaheuristic approaches such as genetic algorithms and evolutionary strategies. The findings highlight growing trends in hybrid quantum–classical integration, the widespread adoption of metaheuristic optimization, and the importance of multi-objective frameworks adapted to quantum hardware constraints. Finally, the review identifies key research gaps and future directions for practical QCNN deployment on near-term quantum devices.

Algorithms doi: 10.3390/a19060489

Authors: Linhui Wang Zhizhuang Liu Yu Yang Lizhi Chen Zhenqi Zhou Mengyu Zeng Yonghong Tan

In UAV-based remote sensing, accurate and efficient image mosaicing is crucial for achieving real-time monitoring. Traditional cloud-centric processing paradigms, however, face core scientific challenges such as high latency, bandwidth bottlenecks, and limited autonomy, making them inadequate for dynamic, real-time scenarios. To address these issues, this paper proposes an edge-computing-enabled UAV image mosaicing system. The system consists of a UAV remote sensing platform and an edge computing terminal, with the core being our novel B-SIFT-ILS algorithm. The algorithm first uses geographic coordinates for unified registration, constructs a Gaussian scale space for multi-resolution representation, and then precisely locates extrema in the Difference of Gaussian (DoG) space using a 3D quadratic function. A BANSAC algorithm is subsequently employed to refine feature points and extract stable SIFT features, and finally, Iterative Least Squares (ILS) are used to achieve seamless mosaicing. Experimental results demonstrate that, compared with classical RANSAC, the proposed method achieves superior feature sampling accuracy (rotation: 0.879, translation: 0.877) and lower latency. The ILS-based smoothing stage effectively eliminates noise and ghosting without introducing gradient reversal, performing comparably to deep learning methods while significantly outperforming direct averaging and Gaussian approaches. On the NVIDIA Jetson Orin NX edge terminal, a single processing instance requires only 1124 ms, highlighting its strong potential for real-time, low-latency, and autonomous mosaicing tasks. Future research will focus on extending the approach to non-planar terrains and implementing adaptive parameter tuning for the BANSAC algorithm.

Algorithms doi: 10.3390/a19060488

Authors: Ojobo Agbo Eje Sayed Mehedi Azim Iman Dehzangi

Artificial Intelligence (AI) shows significant potential across healthcare domains, including advanced diagnostics, clinical decision support, and personalized medicine. Despite these advancements, the opaque ‘black box’ nature of complex AI models necessitates the application of Explainable Artificial Intelligence (XAI) to ensure trust, accountability, interpretability, and regulatory compliance. This study systematically reviews 76 studies published between 2020 and 2025 that have used XAI in healthcare. Our findings show that XAI models such as SHAP and LIME are predominantly used for structured data applications, such as electronic health records, while other XAI models, such as Grad-CAM and Layer-wise Relevance Propagation (LRP), are mainly used in medical imaging. This study specifically investigates evaluation metrics for operationalizing explainability, including faithfulness, trustworthiness, and regulatory compliance, which distinguishes it from prior descriptive reviews. Our analysis shows that while XAI significantly enhances clinician trust, thorough explanation remains heterogeneous and largely confined to controlled settings and the employed benchmark datasets. Critical barriers to clinical adoption include inconsistent interpretability across data modalities and the lack of standardized evaluation frameworks. Existing XAI techniques often do not correspond with strict regulatory standards such as the EU AI Act, Food and Drug Administration (FDA) guidelines, and the Health Insurance Portability and Accountability Act (HIPAA). This review argues for the urgent standardization of XAI validation and the development of human-centered designs to move beyond algorithmic transparency toward reliable real-world hospital integration.

Algorithms doi: 10.3390/a19060487

Authors: Wei Xu Yi Wan T. Zuo

Scheduling in multi-layer industrial networks must remain stable even when the feedback mechanism of the environment changes inside a single production episode. The system can switch between a step-continuous regime with dense process feedback and a task-driven regime with sparse milestone feedback, so that the same state requires different behaviour before and after the switch. A regime-oblivious policy may therefore optimise the wrong action preference after a switch. We formulate this setting as a mode-switched multi-industrial-chain Markov decision process (MS-MIC-MDP) and prove that a single fixed action preference is necessarily suboptimal in at least one regime. We then propose BHERA, a belief-guided homeostatic estimation framework for regime adaptation. BHERA builds cross-layer representations, performs structured variational inference of slow and fast latent beliefs, estimates the posterior probability of the task-driven regime, and uses that posterior to regulate sample weights, entropy strength, return-prediction emphasis, and latent information capacity. A homeostatic feedback rule on the Kullback–Leibler (KL) divergence keeps the latent representation informative without allowing uncontrolled information growth, and we analyse it as a two-timescale stochastic approximation with an associated convergence argument and a per-iteration complexity bound. Experiments in a multi-layer industrial scheduling simulator show that BHERA achieves higher return, lower cost, and higher utility than CReSCENT, HiTAC-MuSE, Informed Switching, and WToE across all tested perturbations, with paired statistical tests confirming significance. Expanded ablations and parameter-sensitivity studies confirm the importance of regime belief, regime-balanced weighting, bootstrap prediction, homeostatic capacity control, and the dual-timescale latent split.

Algorithms doi: 10.3390/a19060486

Authors: Benkaihoul Said Bakria Derradji Ibrahim Farouk Bouguenna Habib Benbouhenni Riyadh Bouddou Yıldırım Özüpak Nasreddine Bouchikhi Alin-Gheorghita Mazare Nicu Bizon

Permanent magnet synchronous motors (PMSMs) are widely used in industrial applications due to their high efficiency, compact structure, and excellent dynamic performance. However, achieving accurate speed control with high robustness under load disturbances and parameter uncertainties remains a significant challenge. Conventional proportional–integral (PI) controllers often suffer from overshoot, slow dynamic response, and sensitivity to nonlinear operating conditions. To address these limitations, this paper proposes an intelligent control strategy that combines third-order sliding mode control (TOSMC) with the Golden Jackal Optimization (GJO) algorithm for optimal PMSM speed regulation. The proposed TOSMC-GJO approach aims to enhance the operational performance, robustness, and reliability of PMSM drives. The control structure consists of an optimized outer-loop speed controller and an inner-loop predictive current controller to improve current quality and eliminate the need for conventional PI tuning. The controller parameters are optimized using a fitness function designed to minimize tracking error, overshoot, settling time, torque ripples, and total harmonic distortion (THD). Simulation results under variable speed and load torque conditions demonstrate that the proposed TOSMC-GJO controller achieves superior performance compared with PI control and TOSMC optimized using Grey Wolf Optimization (GWO). The proposed strategy eliminates speed overshoot and reduces the response time to 0.0052 s, compared with 0.0056 s for TOSMC-GWO and 0.011 s for PI control. In addition, the THD of stator currents is reduced to 6.12%, improving current quality and reducing harmonic distortion. The proposed controller also provides smoother torque response, better disturbance rejection capability, and improved waveform symmetry. These results confirm that integrating high-order nonlinear control with metaheuristic optimization significantly improves the dynamic performance, operational reliability, and robustness of PMSM drive systems under demanding operating conditions.

Algorithms doi: 10.3390/a19060485

Authors: Marco Antonio Márquez-Vera Jorge A. Ruiz-Vanoye Carlos Antonio Márquez-Vera Alfian Ma’arif Edith Mendoza-Ramírez

Evolving fuzzy systems (EFS) offer an incremental learning, making them promising for fault diagnosis (FD) in industrial processes, where unknown faults and changing operation conditions are common. The evolving fuzzy structure enables incremental rule adaptation while maintaining interpretability and reduced computational complexity compared with deep learning approaches. However, the performance of EFS depends heavily on the preprocessing of input data. This study evaluates eight preprocessing strategies for EFS applied to the Tennessee Eastman benchmark process. A one-vs-rest EFS architecture was implemented for ten representative faults (IDV1, IDV2, IDV4, IDV5, IDV6, IDV7, IDV8, IDV10, IDV13 and IDV14) in order to make a comparison with other FD techniques. This approach uses seven variables selected by using the least angle regression. Preprocessing methods were applied to highlight fault signatures. Using the Daubechies-4 in the preprocessing achieved the best overall F1-score (73.68%) with a sensitivity of 97.37%, outperforming the no-preprocessing baseline (F1 = 70.67%). Per-fault analysis showed high performance for faults IDV6, IDV7, and IDV14, while IDV1, IDV2, IDV5, and IDV8 exhibited high sensitivity but lower specificity. These findings indicate that wavelet preprocessing significantly enhances EFS for FD, and that the choice of wavelet should be guided by application priorities: Daubechies-4 is recommended for maximum detection and fewer false alarms. The obtained results demonstrate that wavelet preprocessing substantially improves classification robustness and fault discrimination compared with the non-preprocessed baseline.

Algorithms doi: 10.3390/a19060484

Authors: Wencai He Zhijie Peng Yuanbin He Ziyu Zhang Mingshen Zhang He Zhu

Community detection in attributed networks becomes considerably more challenging when the number of communities is unknown in advance. Most existing deep community detection methods assume a fixed community count, whereas reinforcement learning (RL)-based alternatives often suffer from overestimated action values and unstable target updates. To address these limitations, we propose GCR-Net (Graph Community Recognition Network), an RL-guided framework that combines representation learning with adaptive community-count selection. The method adapts decoupled value estimation and gradual anchor-network updates from established deep RL techniques to a formal MDP over candidate community counts. Experiments on citation, social, biomedical, and proteininteraction benchmarks, together with synthetic graphs with more than ten communities, show that GCR-Net delivers competitive NMI and ARI scores with lower variance and more stable optimization than conventional RL baselines. Statistical tests indicate that the clearest gains concern training stability rather than large accuracy margins.

Algorithms doi: 10.3390/a19060483

Authors: Neeta Gajanan Kadukar Diksha Dani

The proliferation of IoT-based surveillance has caused a sharp rise in video data, straining network bandwidth and cloud storage. Conventional video compression exploits pixel-level redundancy but ignores the semantic importance of content, transmitting large volumes of redundant background. This paper proposes a semantic-aware video offloading framework that improves bandwidth efficiency in cloud-based surveillance. DeepLabV3+ with a ResNet-50 backbone performs semantic segmentation at the edge to extract relevant foreground objects (e.g., pedestrians and vehicles) while suppressing static background. A background reference caching mechanism transmits the static scene once and reuses it at the cloud for full-frame reconstruction, minimizing redundant transmission. On a dataset of 12 surveillance sequences (self-captured videos plus sequences from the CDnet 2014 benchmark), the method achieves up to 74.63% reduction in transmitted data, a 33% improvement in storage efficiency, and a compression ratio of 2.88×, while maintaining an average PSNR of 44.92 dB. Paired t-tests (p<0.001) and sensitivity analysis across varying scene dynamics and semantic configurations confirm the robustness of the approach, and comparisons indicate clear gains over conventional motion-based offloading in bandwidth efficiency and reconstruction fidelity.

Algorithms doi: 10.3390/a19060482

Authors: Linjiang Shang Huanyu Cheng Bo Zhou Yin Zhang

Different mixtures of multimodal training data significantly impact the performance of multimodal large language models, and manually tuning data mixtures is inefficient, computationally expensive, and frequently suboptimal because of complex, nonlinear inter-modal interactions. How to determine data-mixture hyperparameters in an efficient and principled manner becomes the bottleneck for progress in the field. This study establishes a scalable, learnable framework, DMPredictor, that treats multimodal data-mixture design as a regression-based hyperparameter-optimization problem and automates the selection of effective training data mixtures. DMPredictor is trained on data mixture samples derived from hundreds of small proxy models (2M parameters), each of which is trained on 1B tokens sampled using different data mixtures. The framework incorporates alignment-aware smoothing and quality-reweighting, enabling diverse exploration of the multimodal data mixture space while avoiding distribution collapse. DMPredictor produces accurate performance forecasts and identifies nearly optimal data mixtures. The predicted optimal mixture surpasses human-designed baselines on diverse benchmarks, achieving +2.7% on MMMU, +6.4% on TextVQA, and +195.2 on MME. Moreover, the mixture optimization complexity is largely reduced by small proxies and a small number of tokens. The proposed approach offers a robust, computationally efficient pathway for optimizing mixtures of multimodal training data, addressing the critical challenge of training data heterogeneity.

Algorithms doi: 10.3390/a19060481

Authors: Lang Yang Changan Ren Zhangwei Yu Mengya Ma

The selection of locations for logistics distribution centers poses a significant challenge in logistics network planning. Traditional methods often demonstrate limited accuracy in solutions and a tendency to become trapped in local optima when addressing large-scale, multi-constraint location models. To address these shortcomings, this study introduces a firefly algorithm enhanced by genetic mutation strategies (GVFA) to optimize the location of distribution centers. Within the framework of the standard firefly algorithm, we incorporate an adaptive step-size decay mechanism and a mutation operator. The movement step size adjusts dynamically based on iteration counts, while a mutation probability of 5% is implemented to maintain population diversity, effectively reducing the risk of premature convergence. A specialized boundary-handling strategy ensures that the search process remains within the feasible solution space, guiding the population toward the global optimum. Experiments were conducted using latitude–longitude coordinates and logistics demand data from 159 Cainiao Post stations in Hengyang City, resulting in the construction of a location model aimed at minimizing total costs. The findings confirm the efficiency and stability of our method in optimizing distribution center locations, thereby providing a novel intelligent optimization approach for the siting of logistics distribution centers.

Algorithms doi: 10.3390/a19060480

Authors: Xing Li Jiaxin Li Huijun Wang Yue Xie Shujuan Jia Zhijie Dong Zitong Yue Baoquan Ma

Graph neural networks have attracted much attention and performed well in many downstream tasks. However, due to issues such as oversmoothing, existing graph neural networks are limited in their ability to quantitatively exploit higher-order neighborhood information. This paper introduces GAtD (Graph Attention Diffusion Method), which propagates attention to a wider range and aggregates higher-order information. We theoretically analyze the effectiveness of GAtD and demonstrate the convergence and linear complexity. A series of experiments demonstrates that, by combining diffusion and attention mechanisms, our method can effectively capture deep level relationships between nodes.

Algorithms doi: 10.3390/a19060479

Authors: José Flores-Salinas Cecilia Rios-Varillas Freddy Tineo-Córdova Julio Cabrera-Chávez Jesús Cernades-Gómez Juan Villalobos-Solano Sonia Escalante-Huamaní Blanca Laines-Lozano

Although interior-point methods (IPMs) have transformed mathematical programming since 1984, the original projective Karmarkar algorithm is rarely documented step by step on reproducible engineering examples that combine algorithmic transparency with real resource allocation constraints. This article therefore does not propose a new variant of Karmarkar’s algorithm; rather, its scientific contribution is the reproducible MATLAB implementation, canonical-form conversion, and comparative validation of the original projective method against the revised Simplex method and Barnes’ affine scaling variant in two engineering settings. The case studies are (i) the minimum-weight plastic design of a rigid frame with seven candidate plastic hinge locations and six collapse mechanisms and (ii) the optimal allocation of crop patterns in the Caplina Valley (Tacna, Southern Peru), an arid irrigated system with an irrigated command area of 1253 ha, monthly labor availability of 22,239 jornales, and water availability derived from Caplina River discharges at 75% persistence. For Case I, the algorithm reached F = 1.001 in the normalized dual space, which corresponds to F = 4.251 in the original structural objective after applying the scaling factor 17/4; relative to the analytical optimum F* = 4.25, this gives |4.251 − 4.25|/4.25 = 2.4 × 10−4 after 20 iterations. For Case II, the model yielded the maximum net production value of USD 703,135.92, allocating 948.47 ha among 12 crops while satisfying water, labor, market, and land constraints. The double validation confirms the algorithm’s strictly interior trajectory, polynomial-time rationale, and transparent internal parameters (α = 0.7968, ε = 10−8), making the implementation a reproducible benchmark for educational use and for future AI–operations research hybrid solvers in regions with limited access to commercial optimization software.

Algorithms doi: 10.3390/a19060478

Authors: Leonardo Loza-Sandoval Robin F. Conchas Jesus G. Alvarez Gabriel Martinez-Soltero Alma Y. Alanis

Complex networks have become a fundamental paradigm for modeling real-world systems. Synchronization of such networks, particularly under hyperchaotic dynamics, presents a significant control challenge due to the high-dimensional state space and multiple positive Lyapunov exponents. This paper addresses the driver node selection problem in a 4D Hyperchaotic Lorenz complex network, formulating it as a constrained binary optimization task. We evaluate a pool of advanced metaheuristics, including the quantum genetic algorithm (QGA), seahorse optimizer (SHO), and artificial bee colony (ABC), across multiple network experiments conducted over 30 independent runs to guarantee statistical validity. The performance of these solvers is rigorously benchmarked against traditional topological heuristics, a random selection baseline comprising 600 feasible configurations, and verified through Wilcoxon statistical testing. Furthermore, addressing computational sustainability, we introduce a “Green-Artificial Intelligence” architecture based on dual-tier structured query language memoization (SQL-memoization) and provide a detailed runtime comparison evaluating its efficiency. The empirical results indicate that swarm-intelligence methods such as ABC and SHO exhibit robust competitive performance in minimizing synchronization errors while the Green-AI framework consistently and drastically reduces the computation of the repetitive simulations.

Algorithms doi: 10.3390/a19060477

Authors: Hassan M. Attar Ahmed G. Elkafas

The maritime sector accounts for approximately 3% of global greenhouse gas (GHG) emissions and faces binding decarbonization obligations under the International Maritime Organization’s (IMO) Net-Zero Framework and the FuelEU Maritime Regulation. Conventional marine fuels, including very low sulphur fuel oil (VLSFO) and liquefied natural gas (LNG), are insufficient to meet long-term regulatory intensity targets on a well-to-wake (WtW) lifecycle basis, creating an urgent need for credible fuel alternatives. This study investigates ethanol as a primary fuel for marine dual-fuel propulsion systems, assessed across four distinct production pathways, sugar beet, corn, sugarcane, and wheat straw, to determine its full decarbonization potential relative to VLSFO and LNG benchmarks. A simulation-based multi-dimensional evaluation framework is developed and applied, integrating dynamic operational simulation, energy analysis, environmental lifecycle modelling, and regulatory compliance assessment. The framework is calibrated against a high-resolution dataset from an active container ship, with scenario-specific engine data. While ethanol requires 39.1% more fuel mass than VLSFO due to its lower energy density, all four ethanol pathways deliver substantially superior WtW GHG reductions: from 50.2% (corn) to 76.9% (wheat straw), compared with 20.6% for LNG. All ethanol scenarios satisfy FuelEU compliance limits across the 2026–2045 horizon, with wheat straw ethanol achieving a GFI of 22.52 gCO2e/MJ, compliant marginally with the 2040 IMO target. These findings demonstrate that bio-based ethanol, particularly from lignocellulosic feedstocks, is a technically viable and regulatorily superior alternative to LNG for maritime decarbonization, warranting accelerated research into production scale-up and bunkering infrastructure development.

Algorithms doi: 10.3390/a19060476

Authors: Massimiliano Caramia

Many deterministic multi-objective optimization methods generate Pareto outcomes by repeatedly solving scalarized subproblems for different preference or reference vectors. When the number of objectives is m≥3, the resulting samples lie on an (m−1)-dimensional Pareto surface in objective space. For tasks such as visualization, trade-off exploration, interactive decision making, and sensitivity analysis, a finite cloud of non-dominated points may be insufficient; one often needs a continuous surrogate of the Pareto surface together with a quantitative control of its reconstruction error. This paper studies the corresponding outer-loop reconstruction problem: how should new reference vectors be selected so as to reconstruct the Pareto surface to a prescribed uniform accuracy while using as few scalarized solves as possible? We propose Certified Adaptive Triangulation Sampling (CATS), a curvature-aware adaptive triangulation method for reconstructing a Pareto surface from an oracle u↦z(u), u∈Δd, where d=m−1. CATS builds a simplicial mesh over the reference simplex and refines the cell with the largest local interpolation quantity η(τ)=12maxkMτ,kdiam(τ)2, where Mτ,k is an upper bound on the Hessian norm of the kth component of the oracle-induced map over τ. This quantity matches the natural error scale of affine interpolation for C2 maps. The rigorous certified interpretation of CATS applies when the preference-to-Pareto map is single-valued, C2, and equipped with reliable local Hessian-norm upper bounds. If such bounds are replaced by numerical curvature estimates, the same rule can still be used as an adaptive refinement indicator, but the resulting stopping test is not a formal certificate unless those estimates are themselves validated. Under the certified assumptions, we prove that the stopping condition maxτη(τ)≤ε guarantees supu∈Δd∥z(u)−z^(u)∥∞≤ε, and that the oracle complexity of certified simplicial piecewise-affine reconstruction is Θ(ε−d/2). On the rigorously certified core tests, CATS uses 2.7×–3.8× fewer oracle calls than uniform reference-direction sampling and 1.2×–1.6× fewer than an AWS-inspired patch-area refinement rule. Additional benchmark studies, evaluated with the same interpolation quantity as a practical stopping indicator, show the same qualitative advantage, especially on anisotropic and localized surface geometries.

Algorithms doi: 10.3390/a19060475

Authors: Armin R. Mikler Chetan Tiwari Murray Patterson

This work concerns the harmonization of geospatial data to improve linkages between place-based characteristics and health outcomes. Such data are typically available as geographic layers, each representing a distinct attribute (e.g., income or distance to a clinic). Since layers are typically constructed independently, their boundaries tend to be spatially incongruent, which can create inconsistencies and introduce bias. This motivates developing algorithmic approaches for aligning such layers while aiming to preserve spatial integrity. This paper formalizes the problem of aligning k collections of m spatial supports over n spatial units in a d-dimensional Euclidean space such that maximum distortion to any collection is minimized. In the above setting, k is the number of layers; n is an indivisible population unit (e.g., census tract); m denotes supports, which are larger regions aggregating a set of contiguous units in order to capture broader regional patterns or enhance statistical stability; and d=2. It is shown that: (1) the one-dimensional case is solvable in time polynomial in k, m, and n; (2) the two-dimensional case is NP-hard for two collections of two supports each; and (3) a heuristic can be provided for aligning a set of collections in the two-dimensional case, which is of practical importance.

Algorithms doi: 10.3390/a19060474

Authors: Jie Sun Mengchao Pu Dongping Tian Yuyu Fan Qinghao Xu Fang Li Siyu Peng

Particle swarm optimization (PSO) is a simple and efficient metaheuristic algorithm that has been widely applied to solving various practical problems. However, PSO has some inherent limitations, such as a tendency to get trapped in local optima and an imbalance between global exploration and local exploitation. To overcome these challenges, this paper proposes a novel algorithm called the multi-swarm particle swarm optimization algorithm with multi-learning strategy (MPLPSO). First, the entire swarm is randomly partitioned into multiple sub-swarms, each comprising three distinct types of particles, which enables the algorithm to explore multiple potential solutions simultaneously. Next, a pool elite learning strategy combined with a convergence learning mechanism is employed to effectively reduce the risk of premature convergence. Furthermore, an elimination-replacement mechanism is integrated with a hierarchical competition strategy to further enhance the solution accuracy. Extensive experiments conducted on the CEC 2017 and CEC 2022 benchmark test suites demonstrate that the proposed MPLPSO significantly outperforms the classical PSO and several state-of-the-art PSO variants. Additionally, MPLPSO is also applied to the traveling salesman problem, and the experimental results further validate the superior performance and robustness of the proposal.

Algorithms doi: 10.3390/a19060473

Authors: Xiaochen Liu Junsan Zhao Guoping Chen

Hyperspectral images (HSIs) and Light Detection and Ranging (LiDAR) data offer complementary spectral and spatial information and are extensively applied to land cover classification. Nevertheless, current fusion–classification approaches frequently suffer from cross-modal feature entanglement and insufficient exploitation of LiDAR physical priors, particularly the Digital Surface Model (DSM), which limits the interpretability of learned features and restricts classification accuracy. To address these issues, this study presents a Physics-Guided Adaptive Decoupling and Collaborative Enhancement Network (ADCE-Net) that embeds explicit geometric guidance into multimodal feature learning. In ADCE-Net, the DSM serves as an explicit geometric conditioning signal to guide feature decoupling, decomposing input representations into modality-shared semantic features (SSF) and modality-specific discriminative features (MSF), thereby mitigating cross-modal interference at an early stage. Based on this decomposition, an adaptive collaborative enhancement mechanism is designed using bidirectional cross-attention and dynamic gating to achieve context-aware mutual refinement between SSF and MSF, facilitating more effective utilization of cross-modal complementary information. Furthermore, a multi-level collaborative classification architecture is constructed to integrate multi-scale contextual representations, enhancing spatial consistency and boundary delineation. Extensive experiments on three benchmark datasets—Trento, Houston 2013, and Muufl Gulfport—demonstrate that ADCE-Net achieves overall accuracies of 99.69%, 97.37%, and 94.90%, respectively, surpassing multiple representative methods including support vector machines, 3D convolutional neural networks, transformer-based models, and recurrent neural networks. Noticeable improvements are also achieved for minority classes and classes with highly similar spectral signatures. The DSM-driven physics guidance boosts both classification performance and feature interpretability, providing a reliable and explainable paradigm for multimodal remote sensing classification.

Algorithms doi: 10.3390/a19060472

Authors: Qi Dong Haolong Jiang Qinyao Liu Yuan Gao

Reliable mathematical models are essential for high-performance analysis and optimization of complex power and energy systems. However, inherent nonlinearities pose significant challenges to accurate model identification. The Hammerstein model, a typical block oriented nonlinear system, consists of a static nonlinear block followed by a linear dynamic block. This paper investigates the data-driven modeling method for the Hammerstein model and proposes a hierarchical identification strategy that integrates the correlation analysis with the Levenberg–Marquardt algorithm. Unlike traditional methods, this hierarchical algorithm strategy decouples the linear and nonlinear modules to avoid parameter coupling and reduces computational complexity. Simulations on a solid oxide fuel cell system and a real-world wind power system confirm the effectiveness and feasibility of the proposed method. The results demonstrate that the hierarchical identification strategy achieves accurate parameter estimation with satisfactory convergence performance.

Algorithms doi: 10.3390/a19060471

Authors: Fan Yang Lixin Lyu

To improve the reliability and path quality of three-dimensional UAV path planning in complex mountainous environments, this paper proposes an Adaptive Elite Differential Gold Rush Optimizer (AEDGRO). The main novelty of AEDGRO lies in the coordinated integration of three enhancement mechanisms into the original Gold Rush Optimizer: chaotic good-point initialization for improving initial population coverage, adaptive elite differential mining for strengthening exploitation around promising regions, and stagnation-aware Gaussian–Cauchy mutation for escaping local optima. A UAV path-planning model is constructed by considering path length, altitude fluctuation, trajectory smoothness, terrain collision avoidance, threat-region avoidance, and UAV safety clearance. The experimental results on the IEEE CEC2017 benchmark suite show that AEDGRO obtains the best Friedman average ranking of 1.63, outperforming the original GRO with a ranking of 4.80. In the UAV path-planning experiments, AEDGRO achieves the lowest mean fitness value of 235.69 and the smallest standard deviation of 7.55, indicating better path quality and stronger robustness than the compared algorithms. The generated trajectories are smoother and can effectively avoid mountainous terrain and threat regions. These results demonstrate that AEDGRO has clear advantages in global optimization accuracy, convergence stability, and UAV path-planning applicability.

Algorithms doi: 10.3390/a19060470

Authors: David B. Tay

Algorithm unrolling (unfolding) is a process where an existing iterative algorithm is converted into another iterative algorithm, but the mapping in each iteration of the new algorithm can potentially be different. An abstraction is to consider a sequence of mappings Tm, where each mapping potentially acts on a different metric space (Xm,dm). We study the iterates from the sequence of mappings and derive conditions for convergence. The first result is when both the mapping and metric space are different in each iteration. The second result is when all metric spaces are the same, but the mapping is different in each iteration. The second result can be considered as a generalization of the Banach Fixed Point theorem. A concrete practical example is the unrolling of the Iterative Shrinkage–Thresholding Algorithm, which has applications in statistics, machine learning and signal processing. The convergence of this example will be analyzed with the aid of the result established through this work.

Algorithms doi: 10.3390/a19060469

Authors: Lucrecia Llerena Nancy Rodríguez Bertha Vásquez John W. Castro Alexander Herrera

Population aging poses significant challenges to oral communication due to age-related changes in articulation, verbal fluency, and speech pacing, even among older adults without neurodegenerative conditions. Despite advances in voice-based assistive technologies, there remains a lack of integrated engineering solutions that support structured, autonomous speech practice in non-clinical environments. This study proposes a deterministic, rule-based speech evaluation workflow implemented within a hybrid web–mobile assistive system. The workflow integrates audio capture, cloud-based automatic speech recognition (ASR), rule-based pronunciation evaluation, immediate multimodal feedback, and progress monitoring within a unified system architecture. The proposed architecture includes a mobile application for older adults and a web platform for configuration and monitoring by caregivers. A prototyping-oriented methodology was applied, including requirements elicitation, system design, implementation, and usability evaluation using the Thinking Aloud method and the System Usability Scale (SUS). Results showed stable system behavior under controlled evaluation conditions, an average recognition accuracy of 90% during preliminary evaluation sessions, and a response latency of 1.82 s, supporting stable real-time interaction during guided speech exercises. These findings demonstrate the feasibility of the proposed assistive architecture as an accessible and reproducible solution for guided speech support in older adults.