Algorithms

21 pages, 5112 KB

Open AccessArticle

A Scalable Framework with Modified Loop-Based Multi-Initial Simulation and Numerical Algorithm for Classifying Brain-Inspired Nonlinear Dynamics with Stability Analysis

by Haseeba Sajjad, Adil Jhangeer and Lubomír Říha

Algorithms 2025, 18(12), 805; https://doi.org/10.3390/a18120805 - 18 Dec 2025

Viewed by 177

Abstract

The principal problem with the analysis of nonlinear dynamical systems is that it is repetitive and inefficient to simulate every initial condition and parameter configuration individually. This not only raises the cost of computation but also constrains scalability in the exploration of a [...] Read more.

The principal problem with the analysis of nonlinear dynamical systems is that it is repetitive and inefficient to simulate every initial condition and parameter configuration individually. This not only raises the cost of computation but also constrains scalability in the exploration of a large parameter space. To solve this, we restructured and extended the computational framework so that variation in the parameters and initial conditions can be automatically explored in a unified structure. This strategy is implemented in the brain-inspired nonlinear dynamical model that has three parameters and multiple coupling strengths. The framework enables detailed categorization of the system responses through statistical analysis and through eigenvalue-based assessment of the stability by considering multiple initial states of the system. These results reveal clear differences between periodic, divergent, and non-divergent behavior and show the extent to which the strength of the coupling

k_{i j}

can drive transitions to stable periodic behavior under all conditions examined. This method makes the analysis process easier, less redundant, and provides a scalable tool to study nonlinear dynamics. In addition to its computational benefits, the framework provides a general method that can be generalized to models with more parameters or more complicated network structures. Full article

(This article belongs to the Special Issue Recent Advances in Numerical Algorithms and Their Applications)

► Show Figures

Figure 1

18 pages, 614 KB

Open AccessReview

Hash Tables as Engines of Randomness at the Limits of Computation: A Unified Review of Algorithms

by Paul A. Gagniuc and Mihai Togan

Algorithms 2025, 18(12), 804; https://doi.org/10.3390/a18120804 - 18 Dec 2025

Viewed by 547

Abstract

Hash tables embody a paradox of deterministic structure that emerges from controlled randomness. They have evolved from simple associative arrays into algorithmic engines that operate near the physical and probabilistic limits of computation. This review unifies five decades of developments across universal and [...] Read more.

Hash tables embody a paradox of deterministic structure that emerges from controlled randomness. They have evolved from simple associative arrays into algorithmic engines that operate near the physical and probabilistic limits of computation. This review unifies five decades of developments across universal and perfect hashing, collision-resolution strategies, and concurrent and hardware-aware architectures. The synthesis shows that modern hash tables act as thermodynamic regulators of entropy, able to transform stochastic mappings into predictable constant-time access. Recent advances in GPU and NUMA-aware designs, lock-free and persistent variants, and neural or quantum-assisted approaches further expand their capabilities. The analysis presents hash tables as models that evolve order within randomness and expand their relevance from classical computation to quantum and neuromorphic frontiers. Full article

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

► Show Figures

Graphical abstract

24 pages, 4145 KB

Open AccessArticle

An Intelligent SPH Framework Based on Machine-Learned Residual Correction for Elliptic PDEs

by Ammar Qarariyah, Tianhui Yang and Fang Deng

Algorithms 2025, 18(12), 803; https://doi.org/10.3390/a18120803 - 18 Dec 2025

Viewed by 267

Abstract

We present an intelligent, non-intrusive framework to enhance the performance of Symmetric Smoothed Particle Hydrodynamics (SSPH) for elliptic partial differential equations, focusing on the linear and nonlinear Poisson equations. Classical Smoothed Particle Hydrodynamics methods, while meshfree, suffer from discretization errors due to kernel [...] Read more.

We present an intelligent, non-intrusive framework to enhance the performance of Symmetric Smoothed Particle Hydrodynamics (SSPH) for elliptic partial differential equations, focusing on the linear and nonlinear Poisson equations. Classical Smoothed Particle Hydrodynamics methods, while meshfree, suffer from discretization errors due to kernel truncation and irregular particle distributions. To address this, we employ a machine-learning-based residual correction, where a neural network learns the difference between the SSPH solution and a reference solution. The predicted residuals are added to the SSPH solution, yielding a corrected approximation with significantly reduced errors. The method preserves numerical stability and consistency while systematically reducing errors. Numerical results demonstrate that the proposed approach outperforms standard SSPH. Full article

► Show Figures

Figure 1

25 pages, 2085 KB

Open AccessArticle

SPR-RAG: Semantic Parsing Retriever-Enhanced Question Answering for Power Policy

by Yufang Wang, Tongtong Xu and Yihui Zhu

Algorithms 2025, 18(12), 802; https://doi.org/10.3390/a18120802 - 17 Dec 2025

Viewed by 241

Abstract

To address the limitations of Retrieval-Augmented Generation (RAG) systems in handling long policy documents, mitigating information dilution, and reducing hallucinations in engineering-oriented applications, this paper proposes SPR-RAG, a retrieval-augmented framework designed for knowledge-intensive vertical domains such as electric power policy analysis. With practicality [...] Read more.

To address the limitations of Retrieval-Augmented Generation (RAG) systems in handling long policy documents, mitigating information dilution, and reducing hallucinations in engineering-oriented applications, this paper proposes SPR-RAG, a retrieval-augmented framework designed for knowledge-intensive vertical domains such as electric power policy analysis. With practicality and interpretability as core design goals, SPR-RAG introduces a Semantic Parsing Retriever (SPR), which integrates community detection–based entity disambiguation and transforms natural language queries into logical forms for structured querying over a domain knowledge graph, thereby retrieving verifiable triple-based evidence. To further resolve retrieval bias arising from diverse policy writing styles and inconsistencies between user queries and policy text expressions, a question-repository–based indirect retrieval mechanism is developed. By generating and matching latent questions, this module enables more robust retrieval of non-structured contextual evidence. The system then fuses structured and unstructured evidence into a unified dual-source context, providing the generator with an interpretable and reliable grounding signal. Experiments conducted on real electric power policy corpora demonstrate that SPR-RAG achieves 90.01% faithfulness—representing a 5.26% improvement over traditional RAG—and 76.77% context relevance, with a 5.96% gain. These results show that SPR-RAG effectively mitigates hallucinations caused by ambiguous entity names, textual redundancy, and irrelevant retrieved content, thereby improving the verifiability and factual grounding of generated answers. Overall, SPR-RAG demonstrates strong deployability and cross-domain transfer potential through its “Text → Knowledge Graph → RAG” engineering paradigm. The framework provides a practical and generalizable technical blueprint for building high-trust, industry-grade question–answering systems, offering substantial engineering value and real-world applicability. Full article

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

► Show Figures

Figure 1

27 pages, 5166 KB

Open AccessArticle

Divergence Shepherd Feature Optimization-Based Stochastic-Tuned Deep Multilayer Perceptron for Emotional Footprint Identification

by Karthikeyan Jagadeesan and Annapurani Kumarappan

Algorithms 2025, 18(12), 801; https://doi.org/10.3390/a18120801 - 17 Dec 2025

Viewed by 235

Abstract

Emotional Footprint Identification refers to the process of recognizing or understanding the emotional impact that a person, experience, or interaction leaves on others. Emotion Recognition plays an important role in human–computer interaction for identifying emotions such as fear, sadness, anger, happiness, and surprise [...] Read more.

Emotional Footprint Identification refers to the process of recognizing or understanding the emotional impact that a person, experience, or interaction leaves on others. Emotion Recognition plays an important role in human–computer interaction for identifying emotions such as fear, sadness, anger, happiness, and surprise on the human face during the conversation. However, accurate emotional footprint identification plays a crucial role due to the dynamic changes. Conventional deep learning techniques integrate advanced technologies for emotional footprint identification, but challenges in accurately detecting emotions in minimal time. To address these challenges, a novel Divergence Shepherd Feature Optimization-based Stochastic-Tuned Deep Multilayer Perceptron (DSFO-STDMP) is proposed. The proposed DSFO-STDMP model consists of three distinct processes namely data acquisition, feature selection or reduction, and classification. First, the data acquisition phase collects a number of conversation data samples from a dataset to train the model. These conversation samples are given to the Sokal–Sneath Divergence shuffling shepherd optimization to select more important features and remove the others. This optimization process accurately performs the feature reduction process to minimize the emotional footprint identification time. Once the features are selected, classification is carried out using the Rosenthal correlative stochastic-tuned deep multilayer perceptron classifier, which analyzes the correlation score between data samples. Based on this analysis, the system successfully classifies different emotions footprints during the conversations. In the fine-tuning phase, the stochastic gradient method is applied to adjust the weights between layers of deep learning architecture for minimizing errors and improving the model’s accuracy. Experimental evaluations are conducted using various performance metrics, including accuracy, precision, recall, F1 score, and emotional footprint identification time. The quantitative results reveal enhancement in the 95% accuracy, 93% precision, 97% recall and 97% F1 score. Additionally, the DSFO-STDMP minimized the in training time by 35% when compared to traditional techniques. Full article

► Show Figures

Figure 1

29 pages, 4009 KB

Open AccessReview

Analysis and Comparison of Machine Learning-Based Facial Expression Recognition Algorithms

by Yuelong Li, Zhanyi Zhou, Quandong Feng and Hongjun Li

Algorithms 2025, 18(12), 800; https://doi.org/10.3390/a18120800 - 17 Dec 2025

Viewed by 492

Abstract

With the rapid development of artificial intelligence technology, facial expression recognition (FER) has gained increasingly widespread applications in digital human generation, humanoid robotics, mental health, and human–computer dialogue. Typical FER algorithms based on machine learning have been widely studied over the past few [...] Read more.

With the rapid development of artificial intelligence technology, facial expression recognition (FER) has gained increasingly widespread applications in digital human generation, humanoid robotics, mental health, and human–computer dialogue. Typical FER algorithms based on machine learning have been widely studied over the past few decades, which motivated our survey. In this study, we have surveyed the state of the art in FER across two categories: traditional machine learning-based (ML-based) and deep learning-based (DL-based) approaches. Each category is analyzed based on six subcategories. Then, twelve methods, including four ML-based models and eight DL-based models, are compared to evaluate FER performance across four datasets. The experimental results show that in validation sets, the average accuracy of HOG-SVM is 50.12%, which is the best performance for the four ML-based methods; in contrast, Poster has an average accuracy of 75.98%, which is the best result obtained among the eight DL-based methods. The most difficult expression to recognize is contempt, with recognition accuracies of 10.00% and 40.06% for ML-based and DL-based methods, respectively. The accuracy of the ML-based method for identifying neutral expression is the highest at 35.25%; the DL-based method has the highest accuracy in identifying surprise at 69.56%. From the theoretical analysis and comparative experimental results of existing methods, we can see that FER faces challenges, including inaccurate recognition in complex environments and unbalanced data categories, highlighting several future research directions, especially those involving the latest applications of digital humans and large language models. Full article

► Show Figures

Figure 1

1 pages, 121 KB

Open AccessCorrection

Correction: Sanjalawe et al. Reinforcement Learning-Guided Hybrid Metaheuristic for Energy-Aware Load Balancing in Cloud Environments. Algorithms 2025, 18, 715

by Yousef Sanjalawe, Salam Al-E’mari, Budoor Allehyani and Sharif Naser Makhadmeh

Algorithms 2025, 18(12), 799; https://doi.org/10.3390/a18120799 - 17 Dec 2025

Viewed by 146

Abstract

In the original publication [...] Full article

17 pages, 1613 KB

Open AccessArticle

Genetic-Based Lottery Ticket Pruning for Transformers in Sentiment Classification: Realized Through Lottery Sample Selection

by Ryan Bluteau, Robin Gras and Gabriel Peralta

Algorithms 2025, 18(12), 798; https://doi.org/10.3390/a18120798 - 17 Dec 2025

Viewed by 268

Abstract

In the growing field of Natural Language Processing (NLP), transformers have become excessively large, pushing the boundaries of both training and inference compute. Given the size and widespread use of these models, there is now a strong emphasis on improving both training and [...] Read more.

In the growing field of Natural Language Processing (NLP), transformers have become excessively large, pushing the boundaries of both training and inference compute. Given the size and widespread use of these models, there is now a strong emphasis on improving both training and inference efficiency. We propose an approach to reduce the computational requirements of transformers. We specifically tested this approach using BERT for sentiment classification. In particular, we reduced the number of attention heads in the model using the lottery ticket hypothesis and an adapted search strategy from a genetic-based lottery ticket pruning algorithm. This search process removes any need for full-sized model training and additionally reduces the training data by up to 95% through lottery sample selection. We achieve leading results in lossless head pruning with a 70% reduction in heads, and up to a 90% reduction with only a 1% F1 loss allocated. The search process was efficiently performed using 5% of training samples under random selection and was further shown to work with just 0.5% of samples by selecting a diverse set of sample embeddings. Inference time was also improved by up to 47.2%. We plan to generalize this work to Large Language Models (LLMs) and language generation tasks to improve both their training and inference requirements. Full article

(This article belongs to the Special Issue Advances in Algorithms Through Heuristics: Theory, Applications, and Innovations)

► Show Figures

Graphical abstract

19 pages, 376 KB

Open AccessArticle

Net Rural Migration Classification in Colombia Using Supervised Decision Tree Algorithms

by Juan M. Sánchez, Helbert E. Espitia and Cesar L. González

Algorithms 2025, 18(12), 797; https://doi.org/10.3390/a18120797 - 16 Dec 2025

Viewed by 203

Abstract

This study presents a decision tree model-based approach to classify rural net migration across Colombian departments using sociodemographic and economic variables. In the model formulation, immigration is considered the movement of people to a destination area to settle there, while emigration is the [...] Read more.

This study presents a decision tree model-based approach to classify rural net migration across Colombian departments using sociodemographic and economic variables. In the model formulation, immigration is considered the movement of people to a destination area to settle there, while emigration is the movement of people from that specific area to other places. The target variable was defined as a binary category representing positive (when the immigration is greater than emigration) or negative net migration. Four classification models were trained and evaluated: Decision Tree, Random Forest, AdaBoost, and XGBoost. Data were preprocessed using cleaning techniques, categorical variable encoding, and class balance assessment. Model performance was evaluated using various metrics, including accuracy, precision, sensitivity,

F_{1}

score, and the area under the ROC curve. The results show that Random Forest achieves the highest accuracy, precision, sensitivity, and

F_{1}

score in the 10-variable and 15-variable settings, while XGBoost is competitive but not dominant. Furthermore, the importance of the model was analyzed to identify key factors influencing migration patterns. This approach allows for a more precise understanding of regional migration dynamics in Colombia and can serve as a basis for designing informed public policies. Full article

(This article belongs to the Special Issue Algorithms in Data Classification (3rd Edition))

► Show Figures

Figure 1

22 pages, 3886 KB

Open AccessArticle

Impact of Image Enhancement Using Contrast-Limited Adaptive Histogram Equalization (CLAHE), Anisotropic Diffusion, and Histogram Equalization on Spine X-Ray Segmentation with U-Net, Mask R-CNN, and Transfer Learning

by Muhammad Shahrul Zaim Ahmad, Nor Azlina Ab. Aziz, Heng Siong Lim, Anith Khairunnisa Ghazali and ‘Afif Abdul Latiff

Algorithms 2025, 18(12), 796; https://doi.org/10.3390/a18120796 - 16 Dec 2025

Viewed by 496

Abstract

Image segmentation is one of the important applications of deep learning models, such as U-Net and Mask R-CNN, in medical imaging. The image segmentation process enables automated extraction of important information within images, including spine X-rays, saving medical practitioners hours of work. However, [...] Read more.

Image segmentation is one of the important applications of deep learning models, such as U-Net and Mask R-CNN, in medical imaging. The image segmentation process enables automated extraction of important information within images, including spine X-rays, saving medical practitioners hours of work. However, for X-ray images, low contrast and noise may affect the quality of the images and consequently reduce the effectiveness of the deep learning models in providing a robust segmentation. Image enhancement prior to feeding the images to segmentation models can help to overcome the issues caused by the low-quality images. This paper aims to evaluate the effects of three image enhancement methods, namely, the contrast-limited adaptive histogram equalization (CLAHE), histogram equalization (HE), and anisotropic diffusion (AD), for improving image segmentation performance of Mask R-CNN, non-transfer learning Mask R-CNN, and U-Net. The findings show image enhancement methods provide significant improvement to the U-Net, and, interestingly, no noticeable improvement of performance on Mask R-CNN is observed. The application of HE for transfer learning Mask R-CNN achieved the highest Dice score of 0.942 ± 0.001 for binary segmentation. The randomly initialized Mask R-CNN obtains the highest DSC of 0.941 ± 0.002 on the same task. On the other hand, for U-Net, despite the presence of statistically significant change by applying image enhancement methods, the model achieves a maximum Dice score of 0.916 ± 0.003, lower than Mask R-CNN with and without transfer learning. A study on image enhancement methods and recent deep learning algorithms is necessary to better understand the effect of image enhancement techniques using deep learning. Full article

(This article belongs to the Special Issue Machine Learning in Medical Signal and Image Processing (4th Edition))

► Show Figures

Figure 1

34 pages, 9590 KB

Open AccessArticle

Selecting Feature Subsets in Continuous Flow Network Attack Traffic Big Data Using Incremental Frequent Pattern Mining

by Sikha S. Bagui, Andrew Benyacko, Dustin Mink, Subhash C. Bagui and Arijit Bagchi

Algorithms 2025, 18(12), 795; https://doi.org/10.3390/a18120795 - 16 Dec 2025

Viewed by 204

Abstract

This work focuses on finding frequent patterns in continuous flow network traffic Big Data using incremental frequent pattern mining. A newly created Zeek Conn Log MITRE ATT&CK framework labeled dataset, UWF-ZeekData24, generated using the Cyber Range at The University of West Florida, was [...] Read more.

This work focuses on finding frequent patterns in continuous flow network traffic Big Data using incremental frequent pattern mining. A newly created Zeek Conn Log MITRE ATT&CK framework labeled dataset, UWF-ZeekData24, generated using the Cyber Range at The University of West Florida, was used for this study. While FP-Growth is effective for static datasets, its standard implementation does not support incremental mining, which poses challenges for applications involving continuously growing data streams, such as network traffic logs. To overcome this limitation, a staged incremental FP-Growth approach is adopted for this work. The novelty of this work is in showing how incremental FP-Growth can be used efficiently on continuous flow network traffic, or streaming network traffic data, where no rebuild is necessary when new transactions are scanned and integrated. Incremental frequent pattern mining also generates feature subsets that are useful for understanding the nature of the individual attack tactics. Hence, a detailed understanding of the features or feature subsets of the seven different MITRE ATT&CK tactics is also presented. For example, the results indicate that core behavioral rules, such as those involving TCP protocols and service associations, emerge early and remain stable throughout later increments. The incremental FP-Growth framework provides a structured lens through which network behaviors can be observed and compared over time, supporting not only classification but also investigative use cases such as anomaly tracking and technique attribution. And finally, the results of this work, the frequent itemsets, will be useful for intrusion detection machine learning/artificial intelligence algorithms. Full article

► Show Figures

Figure 1

4 pages, 136 KB

Open AccessEditorial

Advanced Research on Machine Learning Algorithms in Bioinformatics

by Roberto Pagliarini and Carla Piazza

Algorithms 2025, 18(12), 794; https://doi.org/10.3390/a18120794 - 16 Dec 2025

Viewed by 422

Abstract

Epigenetic variation and somatic mutations represent molecular components of biodiversity that directly link the genome to the environment [...] Full article

(This article belongs to the Special Issue Advanced Research on Machine Learning Algorithms in Bioinformatics)

38 pages, 4891 KB

Open AccessArticle

Thermonuclear Fusion Based Quantum-Inspired Algorithm for Solving Multiobjective Optimization Problems

by Liliya Demidova and Vladimir Maslennikov

Algorithms 2025, 18(12), 793; https://doi.org/10.3390/a18120793 - 15 Dec 2025

Viewed by 366

Abstract

This paper introduces a novel quantum-inspired algorithm for numerical multiobjective optimization, uniquely integrating the multilevel structure of qudits with principles of controlled thermonuclear fusion. Moving beyond conventional qubit-based approaches, the algorithm leverages the qudit’s higher-dimensional state space to enhance search capabilities. Fusion-inspired dynamics—modeling [...] Read more.

This paper introduces a novel quantum-inspired algorithm for numerical multiobjective optimization, uniquely integrating the multilevel structure of qudits with principles of controlled thermonuclear fusion. Moving beyond conventional qubit-based approaches, the algorithm leverages the qudit’s higher-dimensional state space to enhance search capabilities. Fusion-inspired dynamics—modeling particle interaction, energy release, and plasma cooling—provide a powerful metaheuristic framework for navigating complex, high-dimensional Pareto fronts. A hybrid quantum-classical version of the algorithm is presented, designed to exploit the complementary strengths of both computational paradigms for improved efficiency in solving dynamic multiobjective problems. Experimental evaluation on standard dynamic multiobjective benchmarks demonstrates clear performance advantages. Both the quantum-inspired and hybrid variants consistently outperform leading classical algorithms such as NSGA-III, MOEA/D and GDE3, as well as the quantum-inspired NSGA-III, in key metrics: identifying a greater number of unique non-dominated solutions, ensuring superior uniformity along the Pareto front, maintaining stable convergence across generations, and achieving higher accuracy in approximating the ideal solution. Full article

(This article belongs to the Special Issue Numerical Optimization and Algorithms: 4th Edition)

► Show Figures

Figure 1

27 pages, 1148 KB

Open AccessArticle

LLM-Assisted Financial Fraud Detection with Reinforcement Learning

by Ahmed Djalal Hacini, Mohamed Benabdelouahad, Ishak Abassi, Sohaib Houhou, Aissa Boulmerka and Nadir Farhi

Algorithms 2025, 18(12), 792; https://doi.org/10.3390/a18120792 - 15 Dec 2025

Viewed by 643

Abstract

Effective financial fraud detection requires systems that can interpret complex transaction semantics while dynamically adapting to asymmetric operational costs. We propose a hybrid framework in which a large language model (LLM) serves as an encoder, transforming heterogeneous transaction data into a unified embedding [...] Read more.

Effective financial fraud detection requires systems that can interpret complex transaction semantics while dynamically adapting to asymmetric operational costs. We propose a hybrid framework in which a large language model (LLM) serves as an encoder, transforming heterogeneous transaction data into a unified embedding space. These embeddings define the state representation for a reinforcement learning (RL) agent, which acts as a fraud classifier optimized with business-aligned rewards that heavily penalize false negatives while controlling false positives. We evaluate the approach on two benchmark datasets—European Credit Card Fraud and PaySim—demonstrating that policy-gradient methods, particularly A2C, achieve high recall without sacrificing precision. Critically, our ablation study reveals that this hybrid architecture yields substantial performance gains on semantically rich transaction logs, whereas the advantage diminishes on mathematically compressed, anonymized features. Our results highlight the potential of coupling LLM-driven representations with RL policies for cost-sensitive and adaptive fraud detection. Full article

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

► Show Figures

Figure 1

50 pages, 856 KB

Open AccessArticle

LLM-Driven Big Data Management Across Digital Governance, Marketing, and Accounting: A Spark-Orchestrated Framework

by Aristeidis Karras, Leonidas Theodorakopoulos, Christos Karras, George A. Krimpas, Anastasios Giannaros and Charalampos-Panagiotis Bakalis

Algorithms 2025, 18(12), 791; https://doi.org/10.3390/a18120791 - 15 Dec 2025

Viewed by 745

Abstract

In this work, we present a principled framework for the deployment of Large Language Models (LLMs) in enterprise big data management across digital governance, marketing, and accounting domains. Unlike conventional predictive applications, our approach integrates LLMs as auditable, sector-adaptive components that robustly and [...] Read more.

In this work, we present a principled framework for the deployment of Large Language Models (LLMs) in enterprise big data management across digital governance, marketing, and accounting domains. Unlike conventional predictive applications, our approach integrates LLMs as auditable, sector-adaptive components that robustly and directly enhance data curation, lineage, and regulatory compliance. The study contributes (i) a systematic evaluation of seven LLM-enabled functions—including schema mapping, entity resolution, and document extraction—that directly improve data quality and operational governance; (ii) a distributed architecture that deploys Apache Spark orchestration with Markov Chain Monte Carlo sampling to achieve quantifiable uncertainty and reproducible audit trails; and (iii) a cross-sector analysis demonstrating robust semantic accuracy, compliance management, and explainable outputs suited to diverse assurance requirements. Empirical evaluations reveal that the proposed architecture persistently attains elevated mapping precision, resilient multimodal feature extraction, and consistent human supervision. These characteristics collectively reinforce the integrity, accountability, and transparency of information ecosystems, particularly within compliance-driven organizational settings. Full article

► Show Figures

Figure 1

21 pages, 1667 KB

Open AccessArticle

Advanced Retinal Lesion Segmentation via U-Net with Hybrid Focal–Dice Loss and Automated Ground Truth Generation

by Ahmad Sami Al-Shamayleh, Mohammad Qatawneh and Hany A. Elsalamony

Algorithms 2025, 18(12), 790; https://doi.org/10.3390/a18120790 - 14 Dec 2025

Viewed by 388

Abstract

An early and accurate detection of retinal lesions is imperative to intercept the course of sight-threatening ailments, such as Diabetic Retinopathy (DR) or Age-related Macular Degeneration (AMD). Manual expert annotation of all such lesions would take a long time and would be subject [...] Read more.

An early and accurate detection of retinal lesions is imperative to intercept the course of sight-threatening ailments, such as Diabetic Retinopathy (DR) or Age-related Macular Degeneration (AMD). Manual expert annotation of all such lesions would take a long time and would be subject to interobserver tendencies, especially in large screening projects. This work introduces an end-to-end deep learning pipeline for automated retinal lesion segmentation, tailored to datasets without available expert pixel-level reference annotations. The approach is specifically designed for our needs. A novel multi-stage automated ground truth mask generation method, based on colour space analysis, entropy filtering and morphological operations, and creating reliable pseudo-labels from raw retinal images. These pseudo-labels then serve as the training input for a U-Net architecture, a convolutional encoder–decoder architecture for biomedical image segmentation. To address the inherent class imbalance often encountered in medical imaging, we employ and thoroughly evaluate a novel hybrid loss function combining Focal Loss and Dice Loss. The proposed pipeline was rigorously evaluated on the ‘Eye Image Dataset’ from Kaggle, achieving a state-of-the-art segmentation performance with a Dice Similarity Coefficient of 0.932, Intersection over Union (IoU) of 0.865, Precision of 0.913, and Recall of 0.897. This work demonstrates the feasibility of achieving high-quality retinal lesion segmentation even in resource-constrained environments where extensive expert annotations are unavailable, thus paving the way for more accessible and scalable ophthalmological diagnostic tools. Full article

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

► Show Figures

Figure 1

16 pages, 689 KB

Open AccessArticle

The Impacts of Large Language Model Addiction on University Students’ Mental Health: Gender as a Moderator

by Ibrahim A. Elshaer and Alaa M. S. A. Azazz

Algorithms 2025, 18(12), 789; https://doi.org/10.3390/a18120789 - 12 Dec 2025

Viewed by 467

Abstract

This study tested the impacts of large language model (LLM) addiction on the mental health of university students, employing gender as a moderator. Data was collected from 750 university students from multiple fields of study (i.e., business, medical, education, and social sciences) using [...] Read more.

This study tested the impacts of large language model (LLM) addiction on the mental health of university students, employing gender as a moderator. Data was collected from 750 university students from multiple fields of study (i.e., business, medical, education, and social sciences) using a self-administered questionnaire. Partial Least Squares Structural Equation Modeling (PLS-SEM) was employed to analyze the collected data; this study tested the impacts of three LLM addiction dimensions—withdrawal and health problems (W&HPs), time management and performance (TM&P), and social comfort (SC)—on stress, depression, and anxiety as dimensions of mental health disorders. Findings indicate that TM&P and SC had a significant positive impact on stress, depression, and anxiety, implying that overdependence (as an early-stage precursor and behavioral antecedent of LLM addiction) on LLMs for academic achievements and emotional reassurance contributed to higher levels of psychological distress. On the contrary, W&HP showed a weak but significant negative correlation with stress, signaling a probable self-regulatory coping approach. Furthermore, gender was found to successfully moderate several of the tested relationships, where male university students showed stronger relationships between LLM addiction dimensions and adverse mental health consequences, whereas female university students proved greater emotional constancy and resilience. Theoretically, this paper extends the digital addiction frameworks into the AI setting, highlighting gendered models of emotional exposure. Practically, this study highlights the urgent need for gender-sensitive digital well-being intervention programs that address the overuse of LLMs, a prominent category of generative AI. These outcomes emphasize the significance of balancing technological involvement with mental health protection, determining how LLM usage can specifically contribute to digital addiction and related psychological consequences among university students. Full article

(This article belongs to the Special Issue Artificial Intelligence Algorithms and Generative AI in Education (2nd Edition))

► Show Figures

Figure 1

27 pages, 1221 KB

Open AccessArticle

Optimization of Continuous Flow-Shop Scheduling Considering Due Dates

by Feifeng Zheng, Chunyao Zhang and Ming Liu

Algorithms 2025, 18(12), 788; https://doi.org/10.3390/a18120788 - 12 Dec 2025

Viewed by 283

Abstract

For a no-wait flow shop with continuous-flow characteristics, this study simultaneously considers machine setup times and rated processing speed constraints, aiming to minimize the sum of the maximum completion time and the maximum tardiness. First, lower bounds for the maximum completion time, the [...] Read more.

For a no-wait flow shop with continuous-flow characteristics, this study simultaneously considers machine setup times and rated processing speed constraints, aiming to minimize the sum of the maximum completion time and the maximum tardiness. First, lower bounds for the maximum completion time, the maximum tardiness, and the total objective function are developed. Second, a mixed-integer programming (MIP) model is formulated for the problem, and nonlinear elements are subsequently linearized via time discretization. Due to the computational complexity of the problem, two algorithms are proposed: a heuristic algorithm with fixed machine links and greedy rules (HAFG) and a genetic algorithm based on altering machine combinations (GAAM) for solving large-scale instances. The Earliest Due Date (EDD) rule is used as baselines for algorithmic comparison. To better understand the behaviors of the two algorithms, we observe the two components of the objective function separately. The results show that, compared with the EDD rule and GAAM, the HAFG algorithm tends to focus more on optimizing the maximum completion time. The performance of both algorithms is evaluated using their relative deviations from the developed lower bounds and is compared against the EDD rule. Numerical experiments demonstrate that both HAFG and GAAM significantly outperform the EDD rule. In large-scale instances, the HAFG algorithm achieves a gap of about 4%, while GAAM reaches a gap of about 3%, which is very close to the lower bound. In contrast, the EDD rule shows a deviation of about 10%. Combined with a sensitivity analysis on the number of machines, the proposed framework provides meaningful managerial insights for continuous-flow production environments. Full article

► Show Figures

Figure 1

32 pages, 5708 KB

Open AccessArticle

Affordable Audio Hardware and Artificial Intelligence Can Transform the Dementia Care Pipeline

by Ilyas Potamitis

Algorithms 2025, 18(12), 787; https://doi.org/10.3390/a18120787 - 12 Dec 2025

Viewed by 745

Abstract

Population aging is increasing dementia care demand. We present an audio-driven monitoring pipeline that operates either on mobile phones, microcontroller nodes, or smart television sets. The system combines audio signal processing with AI tools for structured interpretation. Preprocessing includes voice activity detection, speaker [...] Read more.

Population aging is increasing dementia care demand. We present an audio-driven monitoring pipeline that operates either on mobile phones, microcontroller nodes, or smart television sets. The system combines audio signal processing with AI tools for structured interpretation. Preprocessing includes voice activity detection, speaker diarization, automatic speech recognition for dialogs, and speech-emotion recognition. An audio classifier detects home-care–relevant events (cough, cane taps, thuds, knocks, and speech). A large language model integrates transcripts, acoustic features, and a consented household knowledge base to produce a daily caregiver report covering orientation/disorientation (person, place, and time), delusion themes, agitation events, health proxies, and safety flags (e.g., exit seeking and falling). The pipeline targets real-time monitoring in homes and facilities, and it is an adjunct to caregiving, not a diagnostic device. Evaluation focuses on human-in-the-loop review, various audio/speech modalities, and the ability of AI to integrate information and reason. Intended users are low-income households in remote settings where in-person caregiving cannot be secured, enabling remote monitoring support for older adults with dementia. Full article

(This article belongs to the Special Issue AI-Assisted Medical Diagnostics)

► Show Figures

Figure 1

19 pages, 318 KB

Open AccessArticle

Bilevel Modelling of Metabolic Networks for Computational Strain Design

by Beichen Wang, Shouyong Jiang, Shibo Xu and Jichun Li

Algorithms 2025, 18(12), 786; https://doi.org/10.3390/a18120786 - 12 Dec 2025

Viewed by 301

Abstract

Bilevel modelling has been widely applied for the identification of genetic perturbations in metabolic engineering. However, most current approaches are based on a biased assumption that mutant strains always grow optimally. In addition, they are developed based on production rates, which may not [...] Read more.

Bilevel modelling has been widely applied for the identification of genetic perturbations in metabolic engineering. However, most current approaches are based on a biased assumption that mutant strains always grow optimally. In addition, they are developed based on production rates, which may not meet yield requirements imposed on a production strain. This paper propose to design strains via multiobjective bilevel models that account for the tradeoff between cell growth and metabolic adjustments from the wild type strain. The proposed modelling frameworks can be used to identify design strategies that maximise rates and/or yields of target products, termed rate-based and yield-based modelling, respectively. We demonstrate, through in silico production of important chemicals in Escherichia coli, that our modelling approaches can generate growth-coupled designs in terms of rate and/or yield, and yield-based modelling identifies design strategies consistent with existing experimental studies as well as suggesting novel designs, thereby holding great promise for selecting targets for high-performance strain design. An important finding from this work that a growth rate coupled design is not necessarily growth yield coupled and vice versa suggests that growth-coupled designs should be analysed in both rate and yield spaces to determine their theoretical feasibility. Full article

(This article belongs to the Special Issue Multi-Objective and Multi-Level Optimization: Algorithms and Applications (2nd Edition))

► Show Figures

Figure 1

20 pages, 6771 KB

Open AccessArticle

Computation of Bounds for Polynomial Dynamic Systems

by Klaus Röbenack and Daniel Gerbet

Algorithms 2025, 18(12), 785; https://doi.org/10.3390/a18120785 - 12 Dec 2025

Viewed by 281

Abstract

Bounds for positive definite sets such as attractors of dynamic systems are typically characterized by Lyapunov-like functions. These Lyapunov functions and their time derivatives must satisfy certain definiteness conditions, whose verification usually requires considerable experience. If the system and a Lyapunov-like candidate function [...] Read more.

Bounds for positive definite sets such as attractors of dynamic systems are typically characterized by Lyapunov-like functions. These Lyapunov functions and their time derivatives must satisfy certain definiteness conditions, whose verification usually requires considerable experience. If the system and a Lyapunov-like candidate function are polynomial, the definiteness conditions lead to Boolean combinations of polynomial equations and inequalities with quantifiers that can be formally solved using quantifier elimination. Unfortunately, the known algorithms for quantifier elimination require considerable computing power, meaning that many problems cannot be solved within a reasonable amount of time. In this context, it is particularly important to find a suitable mathematical formulation of the problem. This article develops a method that reduces the expected computational effort required for the necessary verification of definiteness conditions. The approach is illustrated using the example of the Chua system with cubic nonlinearity. Full article

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

► Show Figures

Figure 1

23 pages, 688 KB

Open AccessArticle

An Adaptive Memetic Differential Evolution with Virtual Population and Multi-Mutation Strategies for Multimodal Optimization Problems

by Tianyan Ding, Zuling Wang, Qingping Liu, Yongtao Wang and Le Yan

Algorithms 2025, 18(12), 784; https://doi.org/10.3390/a18120784 - 11 Dec 2025

Viewed by 222

Abstract

Multimodal optimization is characterized by the dual imperative of achieving global peak diversity and local precision enhancement for discovered solutions. An adaptive memetic differential evolution is proposed in this work based on the virtual population mechanism and multi-mutation strategy to tackle these problems. [...] Read more.

Multimodal optimization is characterized by the dual imperative of achieving global peak diversity and local precision enhancement for discovered solutions. An adaptive memetic differential evolution is proposed in this work based on the virtual population mechanism and multi-mutation strategy to tackle these problems. Firstly, the virtual population mechanism (VPM) is designed to support the maintenance of population diversity, taking advantage of the distribution of a current population to obtain a virtual population. In this mechanism, the virtual population is used to provide certain requirements for the population evolution, but it does not participate in the evolution operation itself. The multi-mutation strategy (MMS) is further executed on the joint virtual and current populations, with the explicit aim of assigning promising candidates to exploitation tasks and less promising ones to exploration tasks during the creation of offspring. Additionally, a probabilistic local search (PLS) scheme is introduced to enhance the precision of elite solutions. This scheme specifically targets the fittest-and-farthest individuals, effectively addressing solution inaccuracies on the identified peaks. Through comprehensive benchmarking on standard test problems, the proposed algorithm demonstrates performance that is either superior or on par with existing methods, confirming its overall competitiveness. Full article

(This article belongs to the Special Issue AI-Driven Distributed Control & Optimization Algorithms for Networked Systems)

► Show Figures

Figure 1

27 pages, 3030 KB

Open AccessSystematic Review

Statistical and Machine Learning Models for Air Quality: A Systematic Review of Methods and Challenges

by Luzneyda Ballesteros Peinado, Teresa Guarda, Germán Herrera-Vidal, Claudia Minnaard and Jairo R. Coronado-Hernández

Algorithms 2025, 18(12), 783; https://doi.org/10.3390/a18120783 - 11 Dec 2025

Viewed by 643

Abstract

Air quality prediction is a critical challenge amid rising environmental and health risks from pollution. This study conducts a systematic literature review (SLR) to compare traditional statistical models and machine learning (ML) techniques applied to air quality forecasting. Following the PRISMA 2020 protocol, [...] Read more.

Air quality prediction is a critical challenge amid rising environmental and health risks from pollution. This study conducts a systematic literature review (SLR) to compare traditional statistical models and machine learning (ML) techniques applied to air quality forecasting. Following the PRISMA 2020 protocol, 412 peer-reviewed articles (2016–2025) were analyzed using thematic filters and bibliometric tools. Results show a marked shift toward ML methods, particularly in Asia (73.2%), with limited representation from Latin America and Africa. Statistical models focused mainly on MLR (88.6%) and ARIMA (11.4%), while ML approaches (n = 574) included Random Forest, LSTM, and SVM. Only 12% of studies conducted direct comparisons. A total of 1177 predictor variables and 307 performance metrics were systematized, highlighting PM2.5, NO₂, and RMSE. Hybrid models like CNN-LSTM show strong potential but face challenges in implementation and interpretability. This review proposes a consolidated framework to guide future research toward more explainable, adaptive, and context-aware predictive models. Full article

(This article belongs to the Special Issue Artificial Intelligence, Image Processing and Spatial Analytics in Environmental Informatics)

► Show Figures

Graphical abstract

25 pages, 3592 KB

Open AccessArticle

Finite Element Computations on Mobile Devices: Optimization and Numerical Efficiency

by Maya Saade, Rafic Younes and Pascal Lafon

Algorithms 2025, 18(12), 782; https://doi.org/10.3390/a18120782 - 11 Dec 2025

Viewed by 304

Abstract

Smartphones have become increasingly powerful and widespread, enabling complex numerical computations that were once limited to desktop systems. However, implementing high-precision Finite Element Analysis (FEA) on mobile devices remains challenging due to constraints in memory, processing speed, and energy efficiency. This paper presents [...] Read more.

Smartphones have become increasingly powerful and widespread, enabling complex numerical computations that were once limited to desktop systems. However, implementing high-precision Finite Element Analysis (FEA) on mobile devices remains challenging due to constraints in memory, processing speed, and energy efficiency. This paper presents an optimized algorithmic framework for performing FEA on mobile platforms, focusing on the adaptation of meshing and iterative solver strategies to resource-limited environments. Several iterative solvers for large sparse linear systems are compared, and predefined refined meshing techniques are implemented to balance computational cost and accuracy. A two-dimensional bridge model is used to validate the proposed methods and demonstrate their numerical stability and computational efficiency on smartphones. The results confirm the feasibility of executing reliable FEA directly on mobile hardware, highlighting the potential of portable, low-cost devices as platforms for computational mechanics and algorithmic simulation in engineering and education. Full article

► Show Figures

Figure 1

30 pages, 3179 KB

Open AccessArticle

Early Student Risk Detection Using CR-NODE: A Completion-Focused Temporal Approach with Explainable AI

by Abdelkarim Bettahi, Hamid Harroud and Fatima-Zahra Belouadha

Algorithms 2025, 18(12), 781; https://doi.org/10.3390/a18120781 - 11 Dec 2025

Viewed by 342

Abstract

Student dropout prediction remains critical in higher education, where timely identification enables effective interventions. Learning Management Systems (LMSs) capture rich temporal data reflecting student behavioral evolution, yet existing approaches underutilize this sequential information. Traditional machine learning methods aggregate behavioral data into static features, [...] Read more.

Student dropout prediction remains critical in higher education, where timely identification enables effective interventions. Learning Management Systems (LMSs) capture rich temporal data reflecting student behavioral evolution, yet existing approaches underutilize this sequential information. Traditional machine learning methods aggregate behavioral data into static features, discarding dynamic patterns that distinguish successful from at-risk students. While Long Short-Term Memory (LSTM) networks model sequences, they assume discrete time steps and struggle with irregular LMS observation intervals. To address these limitations, we introduce Completion-aware Risk Neural Ordinary Differential Equations (CR-NODE), integrating continuous-time dynamics with completion-focused features for early dropout prediction. CR-NODE employs Neural ODEs to model student behavioral evolution through continuous differential equations, naturally accommodating irregular observation patterns. Additionally, we engineer three completion-focused features: completion rate, early warning score, and engagement variability, derived from root cause analysis. Evaluated on Canvas LMS data from 100,878 enrollments across 89,734 temporal sequences, CR-NODE achieves Macro F1 of 0.8747, significantly outperforming LSTM (0.8123), Extreme Gradient Boosting (XGBoost) (0.8300), and basic Neural ODE (0.8682). McNemar’s test confirms statistical significance (

p < 0.0001

). Cross-dataset validation on the Open University Learning Analytics Dataset (OULAD) demonstrates generalizability, achieving 84.44% accuracy versus state-of-the-art LSTM (83.41%). To support transparent decision-making, SHapley Additive exPlanations (SHAP) analysis reveals completion patterns as the primary prediction drivers. Full article

► Show Figures

Graphical abstract

21 pages, 2073 KB

Open AccessArticle

Development and Evaluation of a Real-Time Home Monitoring Application Utilising Long Short-Term Memory Integrated in a Smartphone

by Abdussalam Salama, Reza Saatchi, Maryam Bagheri, Mahpara Saleem and Muhammad Usman Shad

Algorithms 2025, 18(12), 780; https://doi.org/10.3390/a18120780 - 11 Dec 2025

Viewed by 327

Abstract

A novel real-time home monitoring application was developed that utilises long short-term memory (LSTM) and is integrated in a smartphone. Its personalised LSTM accurately learns to detect abnormal movement patterns. The application locally processes the smartphone’s accelerometery data in the form of a [...] Read more.

A novel real-time home monitoring application was developed that utilises long short-term memory (LSTM) and is integrated in a smartphone. Its personalised LSTM accurately learns to detect abnormal movement patterns. The application locally processes the smartphone’s accelerometery data in the form of a signal magnitude vector (SMV) to analyse and interpret the movement patterns. The LSTM was conceptualised by a univariate time-series regression model. It adaptively updates its training parameters by processing the individual’s last seven days of movement data, thus providing a stable, individualised, and dynamic activity baseline. It then quantifies the normal and abnormal movement patterns by continuously comparing the learnt information against the current accelerometery readings. An abnormal movement pattern, e.g., a fall or an unexpected period of inactivity triggers multi-channel alerts to care givers using SMS and email. The application’s performance was evaluated using the data collected from 25 adult volunteers, aged 40–70 years. By interpreting their movement patterns, the application demonstrated a detection accuracy quantified by the coefficient of determination (R²) = 0.93 and an absolute error of 0.05. This precision highlighted a low false positive rate in a real-world evaluation. The study successfully demonstrated a robust, cost-effective, and privacy-preserving home monitoring technology. Full article

(This article belongs to the Special Issue AI-Assisted Medical Diagnostics)

► Show Figures

Figure 1

30 pages, 15770 KB

Open AccessArticle

A Hybrid Deep Learning Framework for Enhanced Fault Diagnosis in Industrial Robots

by Jun Wu, Yuepeng Zhang, Bo Gao, Linzhong Xia, Xueli Zhu, Hui Wang and Xiongbo Wan

Algorithms 2025, 18(12), 779; https://doi.org/10.3390/a18120779 - 10 Dec 2025

Viewed by 390

Abstract

Predominant fault diagnosis in industrial robots depends on dedicated vibration or acoustics sensors. However, their practical deployment is often limited by installation constraints, susceptibility to environmental noise, and cost considerations. Applying Energy-Based Maintenance (EBM) principles to achieve enhanced fault diagnosis under practical industrial [...] Read more.

Predominant fault diagnosis in industrial robots depends on dedicated vibration or acoustics sensors. However, their practical deployment is often limited by installation constraints, susceptibility to environmental noise, and cost considerations. Applying Energy-Based Maintenance (EBM) principles to achieve enhanced fault diagnosis under practical industrial conditions, we propose a hybrid deep learning framework, the Multi-head Graph Attention Network (MGAT) with Multi-scale CNNBiLSTM Fusion (MGAT-MCNNBiLSTM) for industrial robots. This approach obviates the need for additional dedicated sensors, effectively mitigating associated deployment complexities. The framework embodies four core innovations: (1) Based on the EBM paradigm, motor current is established as the most effective and practical choice for enabling cost-efficient and scalable industrial robot fault diagnosis. A corresponding dataset of motor current has been acquired from industrial robots operating under diverse fault scenarios. (2) An integrated MGAT-MCNNBiLSTM architecture that synergistically models multiscale local features and complex dynamics through its MCNNBiLSTM module while capturing nonlinear interdependencies via MGAT. This comprehensive feature representation enables robust and highly accurate fault detection. (3) The study found that the application of spectral preprocessing techniques yields a marked and statistically significant enhancement in diagnostic performance. A comprehensive and systematic analysis was undertaken to uncover the underlying reasons for this observed performance improvement. (4) To emulate challenging industrial settings and cost-sensitive implementations, noise signal injection was employed to evaluate model robustness in high-electromagnetic-interference environments and low-cost, low-resolution ADC implementations. Experimental validation on real-world industrial robot datasets demonstrates that MGAT-MCNNBiLSTM achieves a superior diagnostic accuracy of 90.7560%. This performance marks a significant absolute improvement of 1.51–8.55% over competing models, including LCNNBiLSTM, SCNNBiLSTM, MCCBiLSTM, GAT, and MGAT. Under challenging noise and low-resolution conditions, the proposed model consistently outperforms CNNBiLSTM variants, GAT, and MGAT with an improvement of 1.37–10.26% and enhanced industrial utility and deployment potential. Full article

(This article belongs to the Special Issue Data-Driven Intelligent Modeling and Optimization Algorithms for Industrial Processes: 3rd Edition)

► Show Figures

Figure 1

26 pages, 3434 KB

Open AccessArticle

EEG-Based Decoding of Neural Mechanisms Underlying Impersonal Pronoun Resolution

by Mengyuan Zhao, Hanqing Wang, Yingyi Qiu, Wenlong Wu, Han Liu, Yilin Chang, Xinlin Shao, Yulin Yang and Zhong Yin

Algorithms 2025, 18(12), 778; https://doi.org/10.3390/a18120778 - 10 Dec 2025

Viewed by 364

Abstract

Pronoun resolution is essential for language comprehension, yet the neural mechanisms underlying this process remain poorly characterized. Here, we investigate the neural dynamics of impersonal pronoun processing using electroencephalography combined with machine learning approaches. We developed a novel experimental paradigm that contrasts impersonal [...] Read more.

Pronoun resolution is essential for language comprehension, yet the neural mechanisms underlying this process remain poorly characterized. Here, we investigate the neural dynamics of impersonal pronoun processing using electroencephalography combined with machine learning approaches. We developed a novel experimental paradigm that contrasts impersonal pronoun resolution with direct nominal reference processing. Using electroencephalography (EEG) recordings and machine learning techniques, including local learning-based clustering feature selection (LLCFS), recursive feature elimination (RFE), and logistic regression (LR), we analyzed neural responses from twenty participants. Our approach revealed differential EEG feature patterns across frontal, temporal, and parietal electrodes within multiple frequency bands during pronoun resolution compared to nominal reference tasks, achieving classification accuracies of 78.52% for subject-dependent and 60.10% for cross-subject validation. Behavioral data revealed longer reaction times and lower accuracy for pronoun resolution compared to nominal reference tasks. Combined with differential EEG patterns, these findings demonstrate that pronoun resolution engages more complex mechanisms of referent selection and verification compared to nominal reference tasks. The results establish potential EEG-based indicators for language processing assessment, offering new directions for cross-linguistic investigations. Full article

(This article belongs to the Special Issue Machine Learning Techniques for Brain Data Analysis Using EEG, EMG or Image Data)

► Show Figures

Figure 1

16 pages, 2030 KB

Open AccessArticle

Chinese Text Readability Assessment Based on the Integration of Visualized Part-of-Speech Information with Linguistic Features

by Chi-Yi Hsieh, Jing-Yan Lin, Chi-Wen Hsieh, Bo-Yuan Huang, Yi-Chi Huang and Yu-Xiang Chen

Algorithms 2025, 18(12), 777; https://doi.org/10.3390/a18120777 - 9 Dec 2025

Viewed by 417

Abstract

The assessment of Chinese text readability plays a significant role in Chinese language education. Due to the intrinsic differences between alphabetic languages and Chinese character representations, the readability assessment becomes more challenging in terms of the language’s inherent complexity in vocabulary, syntax, and [...] Read more.

The assessment of Chinese text readability plays a significant role in Chinese language education. Due to the intrinsic differences between alphabetic languages and Chinese character representations, the readability assessment becomes more challenging in terms of the language’s inherent complexity in vocabulary, syntax, and semantics. The article proposed the conceptual analogy between Chinese readability assessment and music’s rhythm and tempo patterns, in which the syntactic structures of the Chinese sentences could be transformed into an image. The Chinese Knowledge and Information Processing Tagger (CkipTagger) tool developed by Sinica-Taiwan is utilized to decompose the Chinese text into a set of tokens. These tokens are then refined through a user-defined token pool to retain meaningful units. An image with part-of-speech (POS) information will be generated by using the token versus syntax alignment. A discrete cosine transform (DCT) is then applied to extract the temporal characteristics of the text. Moreover, the study integrated four categories: linguistic features–type–token ratio, average sentence length, total word, and difficulty level of vocabulary for the readability assessment. Finally, these features were fed into the Support Vector Machine (SVM) network for the classifications. Furthermore, a bidirectional long short-term memory (Bi-LSTM) network is adopted for quantitative comparisons. In simulation, a total of 774 Chinese texts fitted with Taiwan Benchmarks for the Chinese Language were selected and graded by Chinese language experts, consisting of equal amounts of basic, intermediate, and advanced levels. The finding indicated the proposed POS with the linguistic features work well in the SVM network, and the performance matches with the more complex architectures like the Bi-LSTM network in Chinese readability assessments. Full article

(This article belongs to the Topic Applications of NLP, AI, and ML in Software Engineering)

► Show Figures

Figure 1

25 pages, 3721 KB

Open AccessArticle

Forecasting Fossil Energy Price Dynamics with Deep Learning: Implications for Global Energy Security and Financial Stability

by Bilal Ahmed Memon

Algorithms 2025, 18(12), 776; https://doi.org/10.3390/a18120776 - 9 Dec 2025

Viewed by 391

Abstract

This study investigates the application of advanced deep learning models to forecast fossil energy prices, a critical factor influencing global economic stability. Unlike previous research, this study conducts a comparative analysis of Gated Recurrent Unit (GRU), Recurrent Neural Network (RNN), Bidirectional Long Short-Term [...] Read more.

This study investigates the application of advanced deep learning models to forecast fossil energy prices, a critical factor influencing global economic stability. Unlike previous research, this study conducts a comparative analysis of Gated Recurrent Unit (GRU), Recurrent Neural Network (RNN), Bidirectional Long Short-Term Memory (Bi-LSTM), Long Short-Term Memory (LSTM), Convolutional Neural Network (CNN), and Deep Neural Network (DNN) models. The evaluation metrics employed include Root Mean Squared Error (RMSE) and Mean Absolute Percentage Error (MAPE). The results reveal that recurrent architectures, particularly GRU, LSTM, and Bi-LSTM, consistently outperform feedforward and convolutional models, demonstrating superior ability to capture temporal dependencies and nonlinear dynamics in energy markets. In contrast, the RNN and DNN show relatively weaker generalization capabilities. Additionally, visualizations of actual versus predicted prices for each model further emphasize superior forecasting accuracy of recurrent models. The results highlight the potential of deep learning in enhancing investment and policy decisions. Additionally, the results provide significant implications for policymakers and investors by emphasizing the value of accurate energy price forecasting in mitigating market volatility, improving portfolio management, and supporting evidence-based energy policies. Full article

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

► Show Figures

Figure 1

Journal Menu

Journal Browser

Algorithms, Volume 18, Issue 12 (December 2025) – 76 articles

Further Information

Guidelines

MDPI Initiatives

Follow MDPI