Algorithms

20 pages, 847 KB

Open AccessReview

Closing the Loop in Neuromodulation: A Review of Machine Learning Approaches for EEG-Guided Transcranial Magnetic Stimulation

by Elena Mongiardini and Paolo Belardinelli

Algorithms 2026, 19(4), 323; https://doi.org/10.3390/a19040323 - 21 Apr 2026

Viewed by 2143

Abstract

Transcranial magnetic stimulation (TMS) combined with electroencephalography (EEG) provides a powerful framework to probe and modulate human cortical and corticospinal excitability. In recent years, brain state-dependent EEG–TMS paradigms have gained increasing interest by synchronizing stimulation to ongoing neural activity. However, traditional approaches relying [...] Read more.

Transcranial magnetic stimulation (TMS) combined with electroencephalography (EEG) provides a powerful framework to probe and modulate human cortical and corticospinal excitability. In recent years, brain state-dependent EEG–TMS paradigms have gained increasing interest by synchronizing stimulation to ongoing neural activity. However, traditional approaches relying on single oscillatory features or fixed thresholds have yielded heterogeneous and often inconsistent results, motivating the adoption of machine learning (ML) and artificial intelligence (AI) methods to model brain state in a multivariate, data-driven manner. This review synthesizes current ML and deep learning (DL) approaches aimed at predicting cortical and corticospinal excitability from pre-stimulus EEG. We contextualize these methods within brain state-dependent EEG–TMS frameworks based on oscillatory phase, power, and network-level features, and within evolving definitions of brain state that move beyond local biomarkers toward distributed, large-scale, and dynamically evolving neural representations. The reviewed studies span feature-engineered models, data-driven decoding approaches, and emerging adaptive closed-loop frameworks. Finally, we discuss key methodological challenges, translational barriers, and future directions toward personalized, interpretable, and fully closed-loop neuromodulation systems. Full article

(This article belongs to the Special Issue Machine Learning and Signal Processing for EEG, ECG, EDA, and Other Biosignals)

► Show Figures

Figure 1

24 pages, 2106 KB

Open AccessArticle

A Hybrid Deep Learning Framework for Multi-Symbol Recognition and Positional Decoding of Handwritten Babylonian Numerals

by Loay Alzubaidi, Kheir Eddine Bouazza and Islam Al-Qudah

Algorithms 2026, 19(4), 322; https://doi.org/10.3390/a19040322 - 20 Apr 2026

Viewed by 772

Abstract

The Babylonian numeral system, developed more than four thousand years ago, is one of the earliest known positional number systems, employing a sexagesimal (base-60) structure and a limited set of wedge-shaped symbols. Despite their visual simplicity, Babylonian numerals exhibit substantial structural and positional [...] Read more.

The Babylonian numeral system, developed more than four thousand years ago, is one of the earliest known positional number systems, employing a sexagesimal (base-60) structure and a limited set of wedge-shaped symbols. Despite their visual simplicity, Babylonian numerals exhibit substantial structural and positional complexity, particularly when multiple symbols are combined to represent larger numerical values. This complexity presents significant challenges for modern computational recognition, especially in handwritten and degraded archaeological contexts. Most existing research has focused on the recognition of isolated Babylonian numeral symbols, which does not adequately reflect real inscriptions where numerals typically appear as composite sequences. To address this limitation, this paper proposes a hybrid deep learning framework capable of identifying, interpreting, and computing the decimal values of multi-symbol handwritten Babylonian numerals. Building on prior work in single-symbol recognition, we construct a synthetic yet realistic dataset of composite numeral images by combining handwritten glyphs into sequences of two to four symbols while incorporating natural variations in spacing, alignment, and handwriting style. The proposed framework integrates a Convolutional Neural Network (CNN) for visual feature extraction with optional structural feature fusion, followed by a Support Vector Machine (SVM) classifier for reliable multi-class discrimination. A rule-based positional decoder is then applied to convert recognized symbol sequences into their corresponding decimal values using Babylonian base-60 logic. By combining visual recognition with positional numerical reasoning, the proposed system enables end-to-end interpretation of handwritten Babylonian numeral sequences. To the best of our knowledge, this work represents one of the first approaches to jointly classify, decode, and compute numerical values from multi-symbol handwritten Babylonian numerals, contributing to digital epigraphy, archaeological text analysis, and cultural heritage preservation. Full article

(This article belongs to the Special Issue Artificial Intelligence and Data-Driven Approaches for Cultural Heritage)

► Show Figures

Figure 1

20 pages, 9297 KB

Open AccessArticle

D3QN-Guided Sand Cat Swarm Optimization with Hybrid Exploration for Multi-Objective Cloud Task Scheduling

by Minghao Shao, Ying Guo, Jibin Wang and Hu Zhang

Algorithms 2026, 19(4), 321; https://doi.org/10.3390/a19040321 - 20 Apr 2026

Viewed by 469

Abstract

Task scheduling in cloud computing environments is a complex NP-hard problem that requires maximizing resource utilization while satisfying quality-of-service (QoS) constraints. Traditional meta-heuristic algorithms often become stuck in local optima, while single deep reinforcement learning (DRL) models exhibit instability when exploring large-scale solution [...] Read more.

Task scheduling in cloud computing environments is a complex NP-hard problem that requires maximizing resource utilization while satisfying quality-of-service (QoS) constraints. Traditional meta-heuristic algorithms often become stuck in local optima, while single deep reinforcement learning (DRL) models exhibit instability when exploring large-scale solution spaces. To address this, this paper proposes a hybrid scheduling algorithm based on multi-objective sand cat colony optimization (MoSCO). This algorithm utilizes a D3QN network to extract task features and guide population initialization, followed by a multi-objective Sand Cat Swarm Optimization (SCSO) algorithm for refined local search. Results from 50 independent replicate experiments conducted in a simulated cloud environment, coupled with an analysis of the dynamic convergence process, demonstrate that MoSCO exhibits significant superiority and robustness. Scatter plot convergence analysis further confirms that MoSCO’s knowledge injection mechanism effectively overcomes the blind exploration phase of traditional algorithms and successfully breaks through the local optimum bottleneck in the late iteration stages of single reinforcement learning, achieving higher-quality, denser, and more stable convergence. Furthermore, 3D and 2D Pareto front analyses show that MoSCO generates highly competitive, well-distributed non-dominated solutions, offering flexible trade-off options for conflicting objectives. Compared to PureD3QN, H-SCSO, and NSGA-II, MoSCO exhibits the smallest performance fluctuations in box plots. Specifically, MoSCO elevates the average resource utilization of clusters to 92.20%, while reducing the average maximum Makespan and Tardiness to 528 and 4187, respectively. Experimental data confirm that MoSCO effectively balances global exploration with local exploitation, delivering stable, high-quality solutions for dynamic cloud task scheduling. Full article

(This article belongs to the Special Issue Evolutionary and Swarm Computing for Emerging Applications (2nd Edition))

► Show Figures

Figure 1

22 pages, 1684 KB

Open AccessArticle

Assessment of Distributed PV Hosting Capacity in Distribution Areas Based on Operating Region Analysis

by Xiaofeng Dong, Can Liu, Junting Li, Qiong Zhu, Yuying Wang and Junpeng Zhu

Algorithms 2026, 19(4), 320; https://doi.org/10.3390/a19040320 - 20 Apr 2026

Viewed by 515

Abstract

With the high penetration of distributed photovoltaics (PV) in distribution areas, transformer capacity limits and source–load fluctuations have become key factors constraining PV accommodation. To accurately assess the PV hosting capacity under energy storage regulation, this paper proposes an assessment method based on [...] Read more.

With the high penetration of distributed photovoltaics (PV) in distribution areas, transformer capacity limits and source–load fluctuations have become key factors constraining PV accommodation. To accurately assess the PV hosting capacity under energy storage regulation, this paper proposes an assessment method based on operating region analysis. First, a coordinated operation model for the distribution area is established, incorporating the transformer capacity, energy storage constraints, and power balance. On this basis, the calculation boundaries for the PV hosting capacity are discussed in two scenarios: Model 1 ignores power curve uncertainty, characterizing the geometry of the conventional operating region to find the maximum deterministic hosting capacity (S₁) that keeps the region non-empty. Model 2 introduces box-type uncertainty sets for the source and load, proposes the concept of a “Self-Balanced Operating Region”, and constructs a robust feasibility determination model (f₃) based on a Min–Max–Min structure. To solve this multi-layer nested non-convex model, an iterative algorithm based on duality theory and Benders decomposition is employed to determine the robust hosting capacity under uncertainty (S₂) at the critical point where f₃ shifts from zero to non-zero. Case studies show that source–load uncertainty leads to a significant contraction of the operating region, and the robust hosting capacity under uncertainty requirements is strictly less than the deterministic hosting capacity (S₁ > S₂). This method quantifies the reduction effect of uncertainty on the accommodation capability, providing a theoretical basis for planning high-renewable penetration distribution areas and energy storage configuration. Full article

(This article belongs to the Special Issue Algorithms for Electrical and Electronic Engineering with a Focus on Renewable Energy Sources (2nd Edition))

► Show Figures

Figure 1

22 pages, 957 KB

Open AccessArticle

Strategic Capacity Planning Algorithm for Last-Mile Delivery Under High-Volume Demand Surges

by Didar Yedilkhan, Aidarbek Shalakhmetov, Bakbergen Mendaliyev and Nursultan Khaimuldin

Algorithms 2026, 19(4), 319; https://doi.org/10.3390/a19040319 - 18 Apr 2026

Cited by 1 | Viewed by 721

Abstract

Last-mile delivery companies can face demand surges where large-volume order requests exceed daily courier capacity. In such cases fast and robust feasibility-first planning becomes more practical and valuable than building optimal routes. This paper proposes a hierarchical, computationally feasible decomposition pipeline that produces [...] Read more.

Last-mile delivery companies can face demand surges where large-volume order requests exceed daily courier capacity. In such cases fast and robust feasibility-first planning becomes more practical and valuable than building optimal routes. This paper proposes a hierarchical, computationally feasible decomposition pipeline that produces shift-feasible clusters under a strict shift-duration limit using travel-time-based duration estimates. While decomposition methods for large-scale VRPs are well established, they typically remain oriented toward route-construction quality within a single operational day or toward balancing customer counts, demand, or Euclidean territory partitions. In contrast, the proposed method targets a different decision problem: rapid feasibility-first strategic capacity planning for one-time extreme demand surges, where the primary requirement is to estimate, within seconds, a conservative upper bound on the number of courier shifts under a strict shift-duration limit. When end-to-end latency is evaluated from raw geographic points, including distance-matrix preparation for monolithic baselines, the proposed pipeline becomes 187 to 1315 times faster than matrix-based monolithic optimization on the common benchmark sizes. Methodologically, the contribution lies in combining (i) topology-preserving spatial linearization with a Hilbert Space-Filling Curve, (ii) adaptive greedy microclustering driven by empirical travel-time quantiles, and (iii) lexicographic dynamic-programming merge that minimizes the number of shifts first and total travel time second. This yields a planning-oriented decomposition mechanism that is distinct from classical route-quality-centered hierarchical VRP approaches. Full article

(This article belongs to the Section Combinatorial Optimization, Graph, and Network Algorithms)

► Show Figures

Figure 1

34 pages, 1552 KB

Open AccessEditor’s ChoiceReview

On-Orbit Space AI: Federated, Multi-Agent, and Collaborative Algorithms for Satellite Constellations

by Ziyang Wang

Algorithms 2026, 19(4), 318; https://doi.org/10.3390/a19040318 - 17 Apr 2026

Cited by 2 | Viewed by 2923

Abstract

Satellite constellations are transforming space systems from isolated spacecraft into networked, software-defined platforms capable of on-orbit perception, decision making, and adaptation. Yet many of the existing AI studies remain centered on single-satellite inference, while constellation-scale autonomy introduces fundamentally new algorithmic requirements: learning and [...] Read more.

Satellite constellations are transforming space systems from isolated spacecraft into networked, software-defined platforms capable of on-orbit perception, decision making, and adaptation. Yet many of the existing AI studies remain centered on single-satellite inference, while constellation-scale autonomy introduces fundamentally new algorithmic requirements: learning and coordination under dynamic inter-satellite connectivity, strict SWaP-C limits, radiation-induced faults, non-IID data, concept drift, and safety-critical operational constraints. This survey consolidates the emerging field of on-orbit space AI through three complementary paradigms: (i) federated learning for cross-satellite training, personalization, and secure aggregation; (ii) multi-agent algorithms for cooperative planning, resource allocation, scheduling, formation control, and collision avoidance; and (iii) collaborative sensing and distributed inference for multi-satellite fusion, tracking, split/early-exit inference, and cross-layer co-design with constellation networking. We provide a system-level view and a taxonomy that unifies collaboration architectures, temporal mechanisms, and trust models. Full article

(This article belongs to the Collection Feature Papers on Artificial Intelligence Algorithms and Their Applications)

► Show Figures

Graphical abstract

35 pages, 4669 KB

Open AccessArticle

A Hybrid Physics-Informed ML Framework for Emission and Energy Flow Prediction in a Retrofitted Heavy-Duty Vehicle

by Talha Mujahid, Teresa Donateo and Pietropaolo Morrone

Algorithms 2026, 19(4), 317; https://doi.org/10.3390/a19040317 - 17 Apr 2026

Cited by 1 | Viewed by 898

Abstract

This study introduces a physics-informed machine learning framework for predicting transient emissions and energy variables in a retrofitted heavy-duty diesel vehicle. It merges data-driven modeling with physically derived features for reliable real-world analysis. A Random Forest regressor is trained on a public dataset [...] Read more.

This study introduces a physics-informed machine learning framework for predicting transient emissions and energy variables in a retrofitted heavy-duty diesel vehicle. It merges data-driven modeling with physically derived features for reliable real-world analysis. A Random Forest regressor is trained on a public dataset (26 trips from one instrumented vehicle) to predict CO₂ and NO_x mass rates, exhaust temperature, exhaust mass flow rate, and fuel flow rate from synchronized multi-sensor inputs using past-only, time-lagged features. On held-out trips, exhaust temperature prediction achieves

R^{2}

= 0.9997 and

RMSE

= 0.53 g/s; for CO₂, with

R^{2}

= 0.9985 and

RMSE

= 0.38 g/s, comparable performance is reported for NO_x, exhaust flow, and fuel rate. The trained model is integrated into a simulation framework to enable the evaluation of alternative operating conditions and powertrain configurations. First, the impact of cold-start versus hot-start operation is assessed, showing cumulative emission penalties of up to +28% for CO₂ and +30% for NO_x. Second, the effect of hybridization is investigated by comparing the baseline thermal configuration with a hybrid electric architecture, resulting in estimated reductions of −12.2% in CO₂ and −10.5% in NO_x emissions. This tool excels in high-fidelity emission prediction and system-level energy analysis, aiding advanced powertrain assessments under realistic driving conditions. Full article

(This article belongs to the Special Issue Computational Modeling and Intelligent Simulation of Next-Generation Energy Systems)

► Show Figures

Figure 1

28 pages, 1062 KB

Open AccessArticle

Predicting Enterprise AI Adoption in Europe from Cloud Sophistication, Digital Sales Capabilities, and Enterprise Size

by Cristiana Tudor

Algorithms 2026, 19(4), 316; https://doi.org/10.3390/a19040316 - 17 Apr 2026

Viewed by 1018

Abstract

This paper examines whether broad enterprise AI adoption in Europe is best understood as an isolated technology decision or as the outcome of a wider bundle of digital capabilities. Using harmonized Eurostat data for European enterprises, the analysis builds a repeated cross-section at [...] Read more.

This paper examines whether broad enterprise AI adoption in Europe is best understood as an isolated technology decision or as the outcome of a wider bundle of digital capabilities. Using harmonized Eurostat data for European enterprises, the analysis builds a repeated cross-section at the country–size-class–year level and models high AI adoption with a combination of random forest and elastic-net estimation. The dependent variable captures enterprises using at least one AI technology, while the explanatory set focuses on cloud adoption, cloud CRM, cloud ERP, cloud database hosting, cloud security, cloud software use, e-sales intensity, and enterprise size. The findings reveal a stable predictive structure and consistent classification performance across specifications. Across models, cloud CRM and e-sales emerge as the strongest predictors of high AI adoption, followed by general cloud use and selected data-related cloud capabilities. This ordering remains largely stable in threshold-sensitivity checks based on alternative definitions of high adoption. The pattern also remains visible when country controls are removed, which suggests that the result is not merely a reflection of national heterogeneity. The paper contributes by shifting attention from broad claims about “digital readiness” to a narrower and more operational notion of capability complementarity: AI uptake tends to cluster where firms already possess customer-facing, cloud-based, and commercially digital infrastructures. In that sense, the paper offers a transparent, reproducible, and policy-relevant account of the digital foundations of enterprise AI adoption in Europe. Full article

(This article belongs to the Special Issue AI-Driven Business Analytics Revolution)

► Show Figures

Figure 1

21 pages, 635 KB

Open AccessArticle

Agentic Hallucination Risk Scoring for Medical LLMs via Uncertainty Quantification and Clinical Knowledge Injection

by Mayank Kapadia and Mohammad Masum

Algorithms 2026, 19(4), 315; https://doi.org/10.3390/a19040315 - 17 Apr 2026

Viewed by 1302

Abstract

Large Language Models (LLMs) have witnessed significant adoption across numerous domains since 2020, but their proclivity to hallucinate creates unacceptable dangers in high-risk environments like healthcare, where wrong outputs can directly jeopardize human safety. While present systems focus on pre-generation mitigation strategies, they [...] Read more.

Large Language Models (LLMs) have witnessed significant adoption across numerous domains since 2020, but their proclivity to hallucinate creates unacceptable dangers in high-risk environments like healthcare, where wrong outputs can directly jeopardize human safety. While present systems focus on pre-generation mitigation strategies, they cannot ensure the safety of individual outputs during inference. We provide a post hoc Hallucination Risk Scoring (HRS) methodology that intercepts questionable outputs before they reach patients via an agentic pipeline. Given a medical question, a domain-specific LLM generates an initial response from which five complimentary uncertainty signals are computed, which are then separated into a decision layer that governs escalation and a guidance layer that directs clinical knowledge injection by a GPT. The framework is tested using three biological question-answering datasets of various complexity: PubMedQA-Labeled, PubMedQA-Artificial, and BioASQ Task B. The results show an up to 38% safety increase at the most sensitive threshold configuration, zero deterioration across all experimental configurations enforced by the Revert Baseline method, and complexity-aware escalation rates that scale organically with dataset difficulty. Tunable thresholds allow physicians to calibrate system behavior based on deployment requirements, providing a practical safety–accuracy trade-off. Statistical research finds entropy as the primary uncertainty signal separating escalated from non-escalated situations across all datasets. These findings provide a deployable, interpretable, and configurable post hoc safety paradigm for reliable medical AI implementation. Full article

(This article belongs to the Special Issue Evolution of Algorithms in the Era of Generative AI)

► Show Figures

Figure 1

30 pages, 14474 KB

Open AccessArticle

A Data-Driven Spatiotemporal Feature Fusion Method for Traffic Flow Prediction

by Long Li, Zhiwen Wang and Haoxu Wang

Algorithms 2026, 19(4), 314; https://doi.org/10.3390/a19040314 - 16 Apr 2026

Viewed by 383

Abstract

In response to the current severe traffic congestion issues, highly reliable traffic flow prediction serves as a fundamental prerequisite for optimizing municipal road networks and mitigating systemic vehicular congestion. Aiming to elevate the precision of short-term traffic flow prediction, this paper first addresses [...] Read more.

In response to the current severe traffic congestion issues, highly reliable traffic flow prediction serves as a fundamental prerequisite for optimizing municipal road networks and mitigating systemic vehicular congestion. Aiming to elevate the precision of short-term traffic flow prediction, this paper first addresses the low precision of the Dung Beetle Optimizer (DBO) algorithm by introducing an exponential adaptive weight in the way of position update for the ball-rolling dung beetle, along with incorporating a Cauchy–Gaussian mutation strategy. We propose the Multi-strategy improved Dung Beetle Optimizer (MDBO), which is validated using eight benchmark test functions, demonstrating that MDBO outperforms common optimization algorithms in solution accuracy. Secondly, we adopt a combined prediction model, Traffic Flow Temporal-Spatio Network (TFTSNet), which constructs spatial feature modules and temporal feature modules in parallel fusion. Finally, we achieve short-term traffic flow prediction by optimizing the TFTSNet combined prediction model using MDBO. The experiment evaluated model performance using publicly available traffic flow datasets. The results demonstrate that, compared to other state-of-the-art models, the proposed joint prediction model based on MDBO-optimized TFTSNet achieves substantial enhancements in both prediction precision and generalization capability. Root mean square error (RMSE) decreased by 8.7–35.7%, mean absolute error (MAE) decreased by 6.6–40.0%, and R² reached 0.975, showcasing robust predictive capabilities and engineering reference value. Full article

(This article belongs to the Special Issue Bio-Inspired Algorithms: 2nd Edition)

► Show Figures

Figure 1

18 pages, 2012 KB

Open AccessArticle

Design and Analysis of a Reduced Switched-Capacitor Multilevel Inverter-Fed PMSM Drive for Solar–Battery Electric Vehicles Using Rat Swarm Optimization

by Vijaychandra Joddumahanthi, Ramesh Devarapalli and Łukasz Knypiński

Algorithms 2026, 19(4), 313; https://doi.org/10.3390/a19040313 - 16 Apr 2026

Viewed by 848

Abstract

Solar photovoltaic (PV)-powered electric vehicles (EVs) have gained greater significance in the present-day era of transportation across the globe. This proposed work presents an analysis of a five-level reduced switched-capacitor multilevel inverter (RSC-MLI)-powered permanent magnet synchronous motor (PMSM) drive for solar PV-powered battery [...] Read more.

Solar photovoltaic (PV)-powered electric vehicles (EVs) have gained greater significance in the present-day era of transportation across the globe. This proposed work presents an analysis of a five-level reduced switched-capacitor multilevel inverter (RSC-MLI)-powered permanent magnet synchronous motor (PMSM) drive for solar PV-powered battery vehicles enabled by a rat swarm optimization (RSO) maximum power point tracking (MPPT) control mechanism. The system proposed in this paper integrates solar PV arrays and battery storage systems for efficient power transfer to EVs for propulsion. In order to achieve fast, accurate tracking of the optimal maximum power point, the RSO technique is used. A five-level RSC-MLI is used in this study, which enables boosting the voltage and lowering switching losses in the system. The performance of the PMSM is further analyzed to obtain constant parameters, such as the velocity and torque of the electric vehicle. Full article

(This article belongs to the Special Issue Metaheuristic Algorithms in Optimal Design of Engineering Problems (2nd Edition))

► Show Figures

Figure 1

13 pages, 1485 KB

Open AccessArticle

CAHT: A Constraint-Aware Heterogeneous Transformer for Real-Time Multi-Robot Task Allocation in Warehouse Environments

by Shengshuo Gong and Oleg Varlamov

Algorithms 2026, 19(4), 312; https://doi.org/10.3390/a19040312 - 16 Apr 2026

Cited by 2 | Viewed by 1108

Abstract

The NP-hard coordination of heterogeneous robots for time-windowed warehouse tasks remains challenging: metaheuristics are precise but slow, whereas neural methods cannot handle heterogeneous constraints, leading to infeasible allocations. This paper presents the Constraint-Aware Heterogeneous Transformer (CAHT), a lightweight encoder–decoder architecture that performs end-to-end [...] Read more.

The NP-hard coordination of heterogeneous robots for time-windowed warehouse tasks remains challenging: metaheuristics are precise but slow, whereas neural methods cannot handle heterogeneous constraints, leading to infeasible allocations. This paper presents the Constraint-Aware Heterogeneous Transformer (CAHT), a lightweight encoder–decoder architecture that performs end-to-end task assignment and sequencing in a single forward pass. The central innovation is a dynamic feasibility masking mechanism that enforces capacity and energy constraints directly within the softmax computation, eliminating infeasible allocations at the architectural level. This is complemented by a spatial-bias Transformer encoder and a two-stage supervised–reinforcement learning training paradigm using ALNS-generated labels. Experiments across four problem scales (5–20 robots, 50–200 tasks) demonstrate that CAHT achieves objective values within 7–13% of the ALNS reference while being 29–91× faster (23–104 ms vs. 2–3 s). Constraint violation rates remain below 6%, with time-window satisfaction above 94%. Ablation analysis identifies dynamic masking as the dominant contribution (+213% degradation upon removal), and cross-scale generalization reveals that the optimality gap decreases from 13.0% to 10.7% as the problem scale grows. With only 0.91 M parameters, CAHT occupies a new trade-off point on the Pareto frontier, offering a practical path toward real-time autonomous warehouse coordination. Full article

► Show Figures

Figure 1

17 pages, 1597 KB

Open AccessArticle

Strategic Approach for Enhancing Deep Learning Models

by Oded Koren, Yoav Gvili and Liron J. Friedman

Algorithms 2026, 19(4), 311; https://doi.org/10.3390/a19040311 - 16 Apr 2026

Viewed by 1023

Abstract

Modern large language models have achieved remarkable growth and performance across domains, yet their intense use of resources and high computational costs present challenges to scalability and sustainability. Current attempts to surpass baseline (naïve) AutoDL (Automated Deep Learning) models often rely on complex [...] Read more.

Modern large language models have achieved remarkable growth and performance across domains, yet their intense use of resources and high computational costs present challenges to scalability and sustainability. Current attempts to surpass baseline (naïve) AutoDL (Automated Deep Learning) models often rely on complex manipulations to yield marginal accuracy gains while demanding deep domain knowledge and computational intensity. To address known inefficiencies in computation and implementation, this study proposes a strategic approach for enhancing processing without compromising model accuracy or performance through a simplified, scalable methodology. We present a novel AutoDL weight optimization model method that analyzes the most accurate deep learning starting point and achieves the highest outcomes while considering the additional “presetting” analysis overhead. Using 20 real-world datasets, we conducted experiments across three models, six weight configurations, and ten seeds, totaling 62,400 epochs. In all experiments, the optimized model outperformed baselines, achieving higher accuracy across every dataset while reducing preprocessing to only two epochs per seed. These results demonstrate that minimal preprocessing—limited to two epochs per seed—can substantially lower computational demand while maintaining precision. As the global demand for AI deployment accelerates, this conservation-oriented approach will be critical to sustaining innovation within resource and infrastructural constraints, enabling advances in computational sustainability and responsible AI development, tangible savings across multiple dimensions of resource consumption, and broader access to deep learning technologies. Full article

(This article belongs to the Special Issue AI Applications and Modern Industry)

► Show Figures

Figure 1

40 pages, 7468 KB

Open AccessReview

Traffic Flow Prediction in Intelligent Transportation Systems: A Comprehensive Review of Graph Neural Networks and Hybrid Deep Learning Methods

by Zhenhua Wang, Xinmeng Wang, Lijun Wang, Zheng Wu, Jiangang Hu, Fujiang Yuan and Zhen Tian

Algorithms 2026, 19(4), 310; https://doi.org/10.3390/a19040310 - 16 Apr 2026

Cited by 2 | Viewed by 1932

Abstract

Traffic flow prediction is a key component of Intelligent Transportation Systems (ITS), crucial for alleviating urban congestion, optimizing traffic management, and improving the overall efficiency of road networks. With the rapid growth in vehicle numbers and the increasing complexity of urban traffic patterns, [...] Read more.

Traffic flow prediction is a key component of Intelligent Transportation Systems (ITS), crucial for alleviating urban congestion, optimizing traffic management, and improving the overall efficiency of road networks. With the rapid growth in vehicle numbers and the increasing complexity of urban traffic patterns, accurate short-term traffic flow prediction has become increasingly important. This paper comprehensively reviews the latest advancements in traffic flow prediction methods, focusing on graph neural network (GNN)-based approaches and hybrid deep learning frameworks. First, we introduce the fundamental theoretical foundations, including graph neural networks, deep learning algorithms, heuristic optimization methods, and attention mechanisms. Subsequently, we summarize GNN-based prediction methods into four paradigms: (1) federated learning and privacy-preserving methods, enabling cross-regional collaboration while protecting sensitive data; (2) dynamically adaptive graph structure methods, capturing time-varying spatial dependencies; (3) multi-graph fusion and attention mechanism methods, enhancing feature representations from multiple perspectives; and (4) cross-domain technology integration methods, fusing novel architectures and interdisciplinary technologies. Furthermore, we investigate hybrid methods combining signal decomposition, heuristic optimization, and attention mechanisms with LSTM networks to address challenges related to non-stationarity and model optimization. For each category, we analyzed representative works and summarized their core innovations, strengths, and limitations using a systematic comparative table. Finally, we discussed current challenges, including computational complexity, model interpretability, and generalization ability, and outlined future research directions such as lightweight model design, uncertainty quantification, multimodal data fusion, and integration with traffic control systems. This review provides researchers and practitioners with a systematic understanding of the latest advances in traffic flow prediction and offers guidance for methodological selection and future research. Full article

► Show Figures

Figure 1

18 pages, 4176 KB

Open AccessArticle

An Attention-Enhanced Network for Visual Attitude Estimation

by Lu Liu, Jiahao Duan, Yaoyang Shen, Shihan Wang, Jiale Mao, Wei Liu, Yuyan Guo, Lan Wu, Ming Kong and Hang Yu

Algorithms 2026, 19(4), 309; https://doi.org/10.3390/a19040309 - 15 Apr 2026

Viewed by 378

Abstract

Accurate estimation of object attitude is essential for understanding motion behavior and achieving dynamic tracking. Existing image-based methods often suffer from low efficiency and limited accuracy, while the potential of deep learning has not been fully exploited in this field. To address these [...] Read more.

Accurate estimation of object attitude is essential for understanding motion behavior and achieving dynamic tracking. Existing image-based methods often suffer from low efficiency and limited accuracy, while the potential of deep learning has not been fully exploited in this field. To address these limitations, a lightweight deep learning method for attitude estimation is proposed and validated on spherical particles. A synthetic dataset is generated through VTK-based rendering and automatic annotation, providing large-scale training samples with known Euler angles. An improved MobileNetV1 backbone is developed by integrating Squeeze-and-Excitation blocks, a dual-scale Pyramid Pooling Module, global average pooling, and a regression-oriented multilayer perceptron, which enhances feature extraction and enables direct Euler angle prediction. Experimental results show that the proposed method achieves an average error of 0.308° on synthetic test images. Furthermore, a solid particle was fabricated through 3D printing and physical measurements were conducted, where the network combined with image preprocessing and augmentation achieved an average error of about 0.5° on real images, demonstrating a lightweight and deployment-friendly framework for practical attitude estimation. The results verify the effectiveness of the method and demonstrate its potential for accurate and computationally efficient attitude measurement in applications such as fluid dynamics, industrial inspection, and motion tracking. Full article

► Show Figures

Figure 1

22 pages, 1672 KB

Open AccessArticle

A Hybrid Transformer–Graph Framework for Curriculum Sequencing and Prerequisite Optimization in Computer Science Education with Explainable AI

by Ritika Awasthi, Abhinav Shukla, Ayush Kumar Agrawal, Parul Dubey and R Kanesaraj Ramasamy

Algorithms 2026, 19(4), 308; https://doi.org/10.3390/a19040308 - 14 Apr 2026

Viewed by 879

Abstract

Curriculum redesign in Computer Science and Information Technology has become increasingly complex due to rapid technological advancements, interdisciplinary knowledge requirements, and evolving industry expectations. Recent progress in artificial intelligence, particularly Transformer-based language models, offers new opportunities for data-driven and scalable curriculum analysis. This [...] Read more.

Curriculum redesign in Computer Science and Information Technology has become increasingly complex due to rapid technological advancements, interdisciplinary knowledge requirements, and evolving industry expectations. Recent progress in artificial intelligence, particularly Transformer-based language models, offers new opportunities for data-driven and scalable curriculum analysis. This study utilizes syllabus-level textual datasets collected from multiple universities, comprising structured and unstructured course descriptions across diverse CS and IT programs. The dataset enables semantic representation learning and prerequisite inference while supporting cross-institutional curriculum analysis. We propose a hybrid framework that combines Transformer-based semantic encoding with graph-based prerequisite optimization and constraint-aware curriculum sequencing. The novelty of this work lies in integrating semantic prerequisite discovery, optimization-driven curriculum structuring, and explainable AI within a unified decision-support framework. Experimental results demonstrate that the proposed approach consistently outperforms existing machine learning and deep learning baselines, achieving higher prerequisite prediction accuracy, improved curriculum feasibility, and more coherent course sequencing, thereby offering a scalable and interpretable solution for evidence-based curriculum redesign in higher education. Full article

► Show Figures

Figure 1

19 pages, 2355 KB

Open AccessArticle

Hidden Drivers in Ecological Networks: Detecting Possible Latent Forcing from Time-Series Data

by Menglin Liu and Zhihao Cao

Algorithms 2026, 19(4), 307; https://doi.org/10.3390/a19040307 - 14 Apr 2026

Viewed by 467

Abstract

Ecological interaction networks are often influenced by unobserved upstream forcing, which can make an observed link between two variables appear causal even when it primarily reflects mediation within a deeper causal chain. This challenges the interpretation of trophic control, the identification of dominant [...] Read more.

Ecological interaction networks are often influenced by unobserved upstream forcing, which can make an observed link between two variables appear causal even when it primarily reflects mediation within a deeper causal chain. This challenges the interpretation of trophic control, the identification of dominant ecological drivers, and the design of effective monitoring strategies. Here, we introduce detecting pairwise effect (DPE), a data-driven method based on empirical dynamic modeling and multivariable state-space reconstruction, to assess whether an observed directional link is more consistent with a direct ecological driver or with a mediated effect shaped by latent upstream forcing. Using benchmark dynamical system motifs, we show that DPE can identify conditions under which a strong upstream driver acting on the putative cause may distort pairwise causal interpretation. We further demonstrate its ecological relevance using real-world time series, where DPE reveals diagnostic signals consistent with hidden trophic forcing in a plankton food-chain dataset and possible latent external forcing in soil-temperature depth dynamics. Overall, DPE provides a diagnostic signal that is consistent with the presence of latent upstream influence, thereby improving the interpretability of inferred ecological causal relationships. Full article

► Show Figures

Figure 1

26 pages, 4245 KB

Open AccessArticle

RCS-HFPN-YOLOV11: A New Small Target Detection Model

by Hong Zhang, Runzhen Liu, Zhengqing Zhu and Yu Feng

Algorithms 2026, 19(4), 306; https://doi.org/10.3390/a19040306 - 13 Apr 2026

Viewed by 637

Abstract

Despite over two decades of advancement in object detection, achieving high accuracy for small target detection in practical applications remains an unresolved challenge. This paper proposes a novel small-object detection model to address this issue. The model incorporates three key innovations: first, the [...] Read more.

Despite over two decades of advancement in object detection, achieving high accuracy for small target detection in practical applications remains an unresolved challenge. This paper proposes a novel small-object detection model to address this issue. The model incorporates three key innovations: first, the RCSOSA module, which optimizes feature information transmission through dynamic channel interaction and multi-scale feature coordination; second, the HFPN module, a three-branch multi-scale feature fusion network that integrates local and global features by combining CNN and Transformer architectures to enhance semantic details; and third, the NWD-CIoU loss function, which dynamically adjusts the weights of NWD and CIoU losses based on the training phase. Experimental results on the COCO dataset demonstrate that our model improves detection accuracy by 4% over YOLOv11 and achieves state-of-the-art performance among mainstream models while maintaining a real-time inference speed of no less than 60 FPS. Furthermore, validation on the VisDrone dataset confirms the model’s strong generalization capability. The proposed algorithm significantly enhances small target detection accuracy, effectively mitigating a critical limitation in current practical object detection applications. Full article

(This article belongs to the Special Issue Deep Neural Networks and Optimization Algorithms (2nd Edition))

► Show Figures

Figure 1

20 pages, 2481 KB

Open AccessArticle

In Vitro to In Vivo: Bidirectional and High-Precision Generation of In Vitro and In Vivo Neuronal Spike Data

by Masanori Shimono

Algorithms 2026, 19(4), 305; https://doi.org/10.3390/a19040305 - 13 Apr 2026

Viewed by 1049

Abstract

Translational neuroscience relies on both in vitro slice recordings and in vivo recordings. Their spontaneous population dynamics are observed under decisively different conditions, and across independent experiments, there is typically no clear neuron-to-neuron correspondence. Here, we formulate a one-step-ahead, 1 ms binned, bidirectional [...] Read more.

Translational neuroscience relies on both in vitro slice recordings and in vivo recordings. Their spontaneous population dynamics are observed under decisively different conditions, and across independent experiments, there is typically no clear neuron-to-neuron correspondence. Here, we formulate a one-step-ahead, 1 ms binned, bidirectional transfer task between in vitro and in vivo multineuronal spike trains and provide a standardized evaluation procedure for generation across markedly different recording preparations. We train an autoregressive transformer on 1 ms binned, 128-unit binary sequences and introduce Dice loss to directly optimize spike-event overlap under extreme class imbalance, comparing it with Binary Focal Cross-Entropy (γ = 2.0). Across 12 mouse datasets (6 in vitro HD-MEA sessions and 6 in vivo Neuropixels sessions), the method achieves strong within-domain performance and remains above chance for cross-domain generation (ROC-AUC 0.70 ± 0.09 for in vitro → in vivo; 0.80 ± 0.10 for in vivo → in vitro). Because spike events are rare, we report Precision–Recall curves and PR-AUC alongside ROC-AUC to reflect minority-event quality. The present results should be interpreted as predictive generation under preparation/domain shift rather than as direct evidence of preserved causal biological dynamics; whether the framework also reflects features such as E/I balance or oscillatory structure remains an important question for future validation. To our knowledge, this is the first demonstration of bidirectional, time-resolved generation between unpaired in vitro and in vivo population spike trains without assuming cell correspondence, and the framework can be adapted to other sparse neural event data and related event-based datasets when domain-specific validation criteria are defined. Full article

(This article belongs to the Section Algorithms for Multidisciplinary Applications)

► Show Figures

Graphical abstract

30 pages, 1086 KB

Open AccessArticle

Complex-Valued Orthogonal Unitary Superposition Encoding for Robust Three-Qubit Quantum-Error-Correction-Based Image Transmission

by Udara Jayasinghe and Anil Fernando

Algorithms 2026, 19(4), 304; https://doi.org/10.3390/a19040304 - 13 Apr 2026

Viewed by 727

Abstract

Efficient and reliable transmission of compressed images over noisy channels remains a significant challenge due to the high sensitivity to noise. Quantum communication offers a promising solution by encoding classical information into quantum states; however, these states are still susceptible to noise and [...] Read more.

Efficient and reliable transmission of compressed images over noisy channels remains a significant challenge due to the high sensitivity to noise. Quantum communication offers a promising solution by encoding classical information into quantum states; however, these states are still susceptible to noise and quantum decoherence. To address these limitations, we propose a complex-valued orthogonal unitary superposition (COUS) encoding integrated with a three-qubit quantum error correction (QEC) framework for robust and low-complexity quantum image transmission. The COUS encoding preserves both amplitude and phase information, enhancing reconstruction fidelity while maintaining practical scalability. In the proposed system, images are first compressed using either the joint photographic experts group (JPEG) standard or the high-efficiency image file (HEIF) standard and encoded into quantum states. Quantum channel coding is then applied to protect against quantum noise, followed by COUS encoding prior to transmission. At the receiver, the transmitted data undergoes COUS decoding, quantum error correction, quantum decoding, and source decoding to reconstruct the images. Performance improvements are observed across peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), and universal quality index (UQI) metrics. Simulation results demonstrate that the proposed approach outperforms conventional Hadamard encoding-based three-qubit QEC schemes, achieving maximum channel signal-to-noise ratio (SNR) gains of up to 6 dB, and surpasses bandwidth-equivalent classical communication systems employing polar codes, achieving channel SNR gains of up to 12 dB. These results highlight the potential of the proposed method as a practical solution for high-fidelity quantum image communication, overcoming the limitations of existing approaches. Full article

(This article belongs to the Special Issue Signal Processing, Intelligent Analysis, and Optimization for Communication and Electronic Systems)

► Show Figures

Figure 1

24 pages, 6110 KB

Open AccessArticle

Research on Medical Image Segmentation Based on Frequency-Domain Enhancement and Edge Awareness

by Jiamin Li, Yazhi Liu and Wei Li

Algorithms 2026, 19(4), 303; https://doi.org/10.3390/a19040303 - 12 Apr 2026

Viewed by 975

Abstract

Medical images commonly exhibit low contrast, weak boundaries, and complex textures. In addition, significant semantic differences exist between deep-level semantic features and shallow-level detail features, posing challenges for multi-scale feature fusion in terms of detail preservation and structural consistency. To address these issues, [...] Read more.

Medical images commonly exhibit low contrast, weak boundaries, and complex textures. In addition, significant semantic differences exist between deep-level semantic features and shallow-level detail features, posing challenges for multi-scale feature fusion in terms of detail preservation and structural consistency. To address these issues, a frequency-enhanced and bidirectional feature-guided segmentation network (FBNet) is proposed. The network comprises two core components. The frequency-based enhancement (FBE) module employs the Fast Fourier Transform and applies adaptive modulation to the amplitude spectrum through a content-aware gating mechanism, enhancing detail expression and inter-structural contrast. The Bidirectional Guided Feature Fusion module (BGF) enables bidirectional interaction between shallow and deep features. Additionally, the Structure and Edge Awareness (SEA) module is constructed using directional and variance attention mechanisms to achieve collaborative optimization of structural modeling and edge perception. Experiments on four medical image segmentation datasets show that, compared to the second-best method, FBNet achieves improvements of 2.12, 1.57, 1.37, and 1.56 percentage points on the mIoU metric and 1.54, 1.11, 0.84, and 1.03 percentage points on the mDice metric. Full article

(This article belongs to the Section Evolutionary Algorithms and Machine Learning)

► Show Figures

Figure 1

21 pages, 6473 KB

Open AccessArticle

Prediction and Analysis of Geochemical Concentrations of Valuable Components Using Machine Learning Methods

by Syrym Kasenov, Almas Temirbekov, Oleg Gavrilenko, Bekdaulet Khudaibergen, Nurdaulet Pirimzhanov and Nurlan Temirbekov

Algorithms 2026, 19(4), 302; https://doi.org/10.3390/a19040302 - 12 Apr 2026

Viewed by 890

Abstract

This study presents an integrated approach for predicting the spatial distribution of gold, copper, and lithium concentrations using machine learning, geostatistical methods, and multivariate geospatial data. The problem is formulated as a spatially dependent multivariate regression task, distinguishing it from traditional classification-based mineral [...] Read more.

This study presents an integrated approach for predicting the spatial distribution of gold, copper, and lithium concentrations using machine learning, geostatistical methods, and multivariate geospatial data. The problem is formulated as a spatially dependent multivariate regression task, distinguishing it from traditional classification-based mineral prospectivity approaches. A unified database was developed, incorporating geochemical indicators, geomorphometric terrain parameters, remote sensing data, and spatial coordinates. Correlation analysis with an adaptive threshold was applied to optimize the feature set and improve model robustness. The results show that linear methods are limited in capturing nonlinear relationships, while ensemble methods provide significantly higher predictive accuracy. In some cases, geostatistical methods achieve the best performance, emphasizing the importance of spatial structure. Feature importance analysis indicates that gold prediction is primarily driven by geochemical indicators, spatial coordinates, and terrain characteristics. Results for copper and lithium confirm the general applicability of the proposed approach. Overall, the study demonstrates the effectiveness of combining machine learning and geostatistics for modeling geochemical processes. Full article

► Show Figures

Figure 1

78 pages, 7146 KB

Open AccessArticle

A Physics-Guided Machine Learning Algorithm for Non-Ionizing Femur Fracture Classification from RF Spectral Data

by Prince O. Siaw, Yacine Chahba, Ebenezer Adjei, Ahmad Aldelemy, Salamatu Ibrahim and Raed Abd-Alhameed

Algorithms 2026, 19(4), 301; https://doi.org/10.3390/a19040301 - 12 Apr 2026

Cited by 1 | Viewed by 909

Abstract

This paper presents a physics-guided machine learning algorithm for classifying femur fracture presence and subtype using non-ionising radiofrequency (RF) spectral data. Multi-sensor S-parameter responses were generated from a femur phantom model across 1.0–3.0 GHz, producing 104 specimens representing intact bone and three fracture [...] Read more.

This paper presents a physics-guided machine learning algorithm for classifying femur fracture presence and subtype using non-ionising radiofrequency (RF) spectral data. Multi-sensor S-parameter responses were generated from a femur phantom model across 1.0–3.0 GHz, producing 104 specimens representing intact bone and three fracture geometries. An exploratory, effect-size-driven band-selection algorithm identified a compact discriminative region between 1.74 and 1.90 GHz. Interpretable classifiers, including k-nearest neighbours (KNN), decision trees, linear discriminant analysis, and Naïve Bayes, were evaluated under strict specimen-level hold-out protocols to prevent data leakage. The KNN algorithm achieved 99.3% frame-level accuracy and 100% specimen-level accuracy for binary fracture detection while maintaining strong robustness in multiclass subtype classification, validated through sensor ablation and leave-one-subtype-out testing. The results demonstrate that compact, interpretable algorithms operating on band-limited RF spectra can achieve reliable, radiation-free fracture classification, supporting future development of continuous and edge-deployable monitoring systems. Full article

(This article belongs to the Special Issue AI-Driven Engineering Optimization)

► Show Figures

Graphical abstract

25 pages, 6534 KB

Open AccessArticle

Spectral–Spatial State Space Model with Hybrid Attention for Hyperspectral Image Classification

by Mengdi Cheng, Haixin Sun, Fanlei Meng, Qiuguang Cao and Jingwen Xu

Algorithms 2026, 19(4), 300; https://doi.org/10.3390/a19040300 - 11 Apr 2026

Viewed by 738

Abstract

Hyperspectral image (HSI) classification requires the extraction of discriminative features from high-dimensional spatial–spectral data. While the Mamba architecture has shown promise in long-sequence modeling with linear complexity, its application to HSI remains constrained by two major hurdles: the unidirectional causal scanning which fails [...] Read more.

Hyperspectral image (HSI) classification requires the extraction of discriminative features from high-dimensional spatial–spectral data. While the Mamba architecture has shown promise in long-sequence modeling with linear complexity, its application to HSI remains constrained by two major hurdles: the unidirectional causal scanning which fails to capture non-causal global dependencies, and the serialization-induced loss of two-dimensional spatial topology and local textures. To overcome these limitations, we propose HAMamba, a novel Hybrid Attention State Space Model. HAMamba facilitates deep representation learning through two core components: a Multi-Scale Dynamic Fusion (MSDF) module and a Hybrid Attention Mamba Encoder (HAME). Specifically, the MSDF module augments spatial perception through parallelized feature extraction and dynamically weighted integration. The HAME synergizes a Bidirectional Sequence Scan Mamba (BSSM) to establish global semantic context and a Spatial–Spectral Gated Attention (SSGA) module to refine local structural details. Comprehensive experiments on four public benchmark datasets demonstrate that the proposed HAMamba significantly outperforms state-of-the-art approaches, achieving a superior balance between classification accuracy and computational efficiency. Full article

(This article belongs to the Section Evolutionary Algorithms and Machine Learning)

► Show Figures

Figure 1

21 pages, 8142 KB

Open AccessArticle

Robust Deep Learning for Multiclass Power System Fault Diagnosis Using Edge Deployment

by Rakesh Sahu, Pratap Kumar Panigrahi, Deepak Kumar Lal, Rudranarayan Pradhan and Chandrakanta Mahanty

Algorithms 2026, 19(4), 299; https://doi.org/10.3390/a19040299 - 11 Apr 2026

Viewed by 1190

Abstract

This article introduces an intelligent framework using deep learning to recognize and classify different faults through the real-time detection of multiple faults in power distribution systems. A collection of data representing normal operating conditions, alongside various fault scenarios including line-to-ground (LG), line-to-line (LL), [...] Read more.

This article introduces an intelligent framework using deep learning to recognize and classify different faults through the real-time detection of multiple faults in power distribution systems. A collection of data representing normal operating conditions, alongside various fault scenarios including line-to-ground (LG), line-to-line (LL), double line-to-ground (LLG), and three-phase line (LLL) faults, was created using three phase current signals obtained from the Real-Time Digital Simulator (RTDS) microgrid test system. To properly model the system dynamics, a feature extraction method that integrates phase currents, differential currents, summation currents and magnitude results was developed. The temporal features of the fault signals were identified by using a sliding window approach to fit the data. A one-dimensional convolutional neural network (CNN) was developed to identify different types of faults. This model performed well, obtaining nearly 96.15% accuracy while testing. In order to evaluate the feasibility of the approach, the trained model was loaded on Raspberry Pi 5, NodeMCU, ESP32 and existing sensing devices. The fault classification performed in real-time was time-sensitive. The proposed intelligent framework is applicable to low-scale operation for smart grid fault monitoring and protection and it is an economically viable solution. Full article

(This article belongs to the Special Issue AI-Powered Predictive Maintenance: Transforming Industrial Operations Through Intelligent Fault Diagnosis (2nd Edition))

► Show Figures

Figure 1

35 pages, 11189 KB

Open AccessArticle

LUMINA-Net: Acute Lymphocytic Leukemia Subtype Classification via Interpretable Convolution Neural Network Based on Wavelet and Attention Mechanisms

by Omneya Attallah

Algorithms 2026, 19(4), 298; https://doi.org/10.3390/a19040298 - 10 Apr 2026

Viewed by 846

Abstract

Acute Lymphoblastic Leukemia (ALL) is a highly prevalent hematological malignancy, especially in children, for whom precise and prompt subtype identification is essential to establish suitable treatment protocols. Current deep learning-based computer-aided diagnosis (CAD) methods for identifying ALL are hindered by numerous drawbacks, such [...] Read more.

Acute Lymphoblastic Leukemia (ALL) is a highly prevalent hematological malignancy, especially in children, for whom precise and prompt subtype identification is essential to establish suitable treatment protocols. Current deep learning-based computer-aided diagnosis (CAD) methods for identifying ALL are hindered by numerous drawbacks, such as a dependence on solely spatial feature depictions, elevated feature dimensions, computationally extensive deep learning architectures, inadequate multi-layer feature utilization, and poor interpretability. This paper introduces LUMINA-Net, a custom, lightweight, and interpretable deep learning CAD for the automated identification and subtype diagnosis of ALL using microscopic blood smear pictures. LUMINA-Net makes four principal contributions: first, it integrates a self-attention module within a lightweight custom Convolution Neural Network (CNN) to effectively capture long-range spatial relationships across clinically pertinent cytological patterns while preserving a compact design. Second, it employs a Discrete Wavelet Transform (DWT)-based wavelet pooling layer that decreases feature dimensions by up to 96.875% while enhancing the obtained depictions with spatial-spectral information. Third, it utilizes a multi-layer feature fusion strategy that combines wavelet-pooled features from two deep layers with a third fully connected layer to create a discriminating multi-scale feature vector. Fourth, it incorporates Gradient-weighted Class Activation Mapping as a dedicated explainability process to furnish clinicians with apparent visual explanations for each classification decision. Withoit the need for image enhancement or segmentation preprocessing, LUMINA-Net outperforms the competing state-of-the-art methods on the same dataset, achieving a peak accuracy of 99.51%, specificity of 99.84%, and sensitivity of 99.51% on the publicly available Kaggle ALL dataset. This demonstrates that LUMINA-Net has the potential to be a dependable, effective, and clinically interpretable CAD tool for ALL diagnosis. Full article

(This article belongs to the Special Issue Computational, Emerging, and Intelligent Algorithms for Biomedical and Healthcare Applications)

► Show Figures

Graphical abstract

40 pages, 3974 KB

Open AccessArticle

Particle Swarm Optimization Based on Cubic Chaotic Mapping and Random Differential Mutation

by Xingrui Li and Ying Guo

Algorithms 2026, 19(4), 297; https://doi.org/10.3390/a19040297 - 10 Apr 2026

Cited by 1 | Viewed by 792

Abstract

Particle swarm optimization is a metaheuristic optimization algorithm that boasts advantages such as fast convergence speed, fewer tunable parameters, and a simple search mechanism. However, it suffers from premature convergence and insufficient later-stage exploitation, limiting its performance on multimodal and high-dimensional problems. In [...] Read more.

Particle swarm optimization is a metaheuristic optimization algorithm that boasts advantages such as fast convergence speed, fewer tunable parameters, and a simple search mechanism. However, it suffers from premature convergence and insufficient later-stage exploitation, limiting its performance on multimodal and high-dimensional problems. In light of this, this paper proposes a Chaos-based Differential Mutation Particle Swarm Optimization (CDMPSO) algorithm to address these limitations. The algorithm employs four synergistic strategies: cubic chaotic mapping with inverse learning for population initialization; adaptive inertia weight to balance exploration and exploitation; convex lens imaging inverse learning to escape local optima; and random differential mutation to maintain population diversity. Ablation experiments validate the contribution of each strategy, with adaptive weight being the most significant. Comparative experiments demonstrate that CDMPSO achieves an average ranking of 1.00, outperforming standard PSO, CPSO (Constriction Particle Swarm Optimization), ACPSO (Adaptive Chaotic Particle Swarm Optimization), and HPSOALS (Hybrid Particle Swarm Optimization with Adaptive Learning Strategy). On the unimodal function f₁, it attains ultra-high precision of 7.07 × 10⁻²⁴⁸, and on the multimodal function f₉, it uniquely converges to the theoretical optimum of zero. The results demonstrate that CDMPSO possesses excellent convergence precision, global search capability, and robustness, providing an effective solution for complex engineering optimization problems. Full article

(This article belongs to the Section Evolutionary Algorithms and Machine Learning)

► Show Figures

Figure 1

22 pages, 2255 KB

Open AccessArticle

Distributed Stochastic Multi-GPU Hyperparameter Optimization for Transfer Learning-Based Vehicle Detection Under Degraded Visual Conditions

by Zhi-Ren Tsai and Jeffrey J. P. Tsai

Algorithms 2026, 19(4), 296; https://doi.org/10.3390/a19040296 - 10 Apr 2026

Viewed by 759

Abstract

Robust vehicle detection in real-world traffic surveillance remains challenging due to degraded imagery caused by motion blur, adverse weather, and low illumination, which significantly increases detector sensitivity to hyperparameter configurations. This study proposes a “Frugal AI” distributed multi-GPU framework that optimizes hyperparameters via [...] Read more.

Robust vehicle detection in real-world traffic surveillance remains challenging due to degraded imagery caused by motion blur, adverse weather, and low illumination, which significantly increases detector sensitivity to hyperparameter configurations. This study proposes a “Frugal AI” distributed multi-GPU framework that optimizes hyperparameters via a stochastic simplex-based search coupled with five-fold cross-validation. Utilizing three low-cost NVIDIA GTX 1050 Ti GPUs, the framework performs parallel candidate exploration with an asynchronous model-level exchange mechanism to escape local optima without the overhead of gradient synchronization. Seven CNN backbones—VGG16, VGG19, GoogLeNet, MobileNetV2, ResNet18, ResNet50, and ResNet101—were evaluated within YOLOv2 and Faster R-CNN detectors. To address memory constraints (4 GB VRAM), YOLOv2 was selected for extensive benchmarking. Performance was measured using a harmonic precision–recall-based cost metric to strictly penalize imbalanced outcomes. Experimental results demonstrate that under identical wall-clock time budgets, the proposed framework achieves an average 1.38% reduction in aggregated cost across all models, with the highly sensitive VGG19 backbone showing a 4.00% improvement. Benchmarking against Bayesian optimization, genetic algorithms, and random search confirms that our method achieves superior optimization quality with statistical significance (

p < 0.05

). Under a rigorous IoU = 0.75 threshold, the optimized models consistently yielded F1-scores 0.8444

\pm

0.0346. Ablation studies further validate that the collaborative model exchange is essential for accelerating convergence in rugged loss landscapes. This research offers a practical, scalable, and cost-efficient solution for deploying robust AI surveillance in resource-constrained smart city infrastructure. Full article

(This article belongs to the Special Issue Advances in Deep Learning-Based Data Analysis)

► Show Figures

Figure 1

33 pages, 2162 KB

Open AccessArticle

Hybrid Narwhale Optimization with Super Modified Simplex and Runge–Kutta Enhancements: Benchmark Validation and Application to Fuzzy Aggregate Production Planning

by Pasura Aungkulanon, Anucha Hirunwat, Roberto Montemanni and Pongchanun Luangpaiboon

Algorithms 2026, 19(4), 295; https://doi.org/10.3390/a19040295 - 9 Apr 2026

Viewed by 492

Abstract

Aggregate production planning (APP) helps medium-term production, manpower, inventory, and subcontracting decisions match expected demand. Deterministic planning models are generally ineffective in manufacturing due to demand and operational variability. Fuzzy linear programming (FLP) has been frequently used to describe imprecision using membership functions [...] Read more.

Aggregate production planning (APP) helps medium-term production, manpower, inventory, and subcontracting decisions match expected demand. Deterministic planning models are generally ineffective in manufacturing due to demand and operational variability. Fuzzy linear programming (FLP) has been frequently used to describe imprecision using membership functions and satisfaction levels. Despite its versatility, accurate approaches for solving multi-objective FLP-based APP models become computationally expensive as issue size and complexity increase. Thus, metaheuristic algorithms are widely used, although many still have premature convergence, parameter sensitivity, and restricted scalability. This study investigates the Narwhal Optimization Algorithm (NO) as a population-based metaheuristic framework. It proposes two hybrid variants to improve convergence reliability and constraint-handling capability: NO combined with the Super Modified Simplex Method (SMS) for local refinement and NO integrated with a Runge–Kutta-based optimizer (RK) for search stability. These hybrid techniques are tested for solution quality, convergence behavior, and robustness using eight response-surface benchmark functions and four constrained optimization problems. A real-parameter fuzzy APP problem with three goods and a six-month planning horizon uses the best variations. The Elevator Kinematic Optimization (EKO) algorithm, chosen for its compliance with the same mathematical framework and consistent parameter values, is used to compare the offered solutions fairly and controlled. Fuzzy programming uses a max–min satisfaction framework with linear membership functions from positive and negative ideal solutions. Computational experiments assess solution quality, stability, and efficiency for nominal and ±10% demand disturbances. The hybrid NO variants better resist premature convergence, stabilize solutions, and satisfy users more than the original NO and benchmark approaches. For small and medium-sized organizations in dynamic situations, hybrid narwhal-based optimization appears to be a reliable and scalable decision-support solution for APP problems under uncertainty. Full article

(This article belongs to the Special Issue Optimizing Logistics Activities: Models and Applications)

► Show Figures

Figure 1

20 pages, 1820 KB

Open AccessArticle

ID-MSNet: An Enhanced Multi-Scale Network with Convolutional Attention for Pixel-Level Steel Defect Segmentation

by Mohammadreza Saberironaghi, Jing Ren and Alireza Saberironaghi

Algorithms 2026, 19(4), 294; https://doi.org/10.3390/a19040294 - 9 Apr 2026

Cited by 1 | Viewed by 639

Abstract

Automated pixel-level detection of steel surface defects is a critical challenge in manufacturing quality control, complicated by the variation in defect size and shape, low contrast with background textures, and the diversity of defect patterns. This paper proposes ID-MSNet, an enhanced version of [...] Read more.

Automated pixel-level detection of steel surface defects is a critical challenge in manufacturing quality control, complicated by the variation in defect size and shape, low contrast with background textures, and the diversity of defect patterns. This paper proposes ID-MSNet, an enhanced version of the UNet3+ architecture, designed specifically for the segmentation of three common steel surface defect types: inclusions, patches, and scratches. The proposed architecture introduces three targeted modifications: (1) a multi-scale feature learning module (MSFLM) in the encoder that uses dilated convolutions at multiple rates to capture contextual features across different scales, combined with DropBlock regularization and batch normalization to improve generalization; (2) an improved down-sampling (IDS) module that replaces standard max-pooling with learnable strided convolutions fused via 1 × 1 convolution, preserving richer feature representations; and (3) a convolutional block attention module (CBAM) integrated into the skip connections to selectively focus the model on spatially and channel-wise relevant defect regions. Experiments on the publicly available SD-saliency-900 dataset demonstrate that ID-MSNet achieved an 86.19% mIoU, outperforming all compared state-of-the-art segmentation models while using only 6.7 million parameters—approximately 75% fewer than the original UNet3+. These results establish ID-MSNet as a strong and efficient baseline for steel surface defect segmentation, with potential applicability to automated quality inspection in broader manufacturing contexts. Full article

► Show Figures

Figure 1

Journal Menu

Journal Browser

Algorithms, Volume 19, Issue 4 (April 2026) – 80 articles

Further Information

Guidelines

MDPI Initiatives

Follow MDPI