Mohd Nor Akmal Khalid
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Mohd Nor Akmal Khalid has a total of 81 co-authors. Here are the top co-authors by number of shared publications:
- Hiroyuki Iida
School of Information Science, Japan Advanced Institute of Science and Technology, Nomi, Ishikawa, Japan
34 shared publications - Umi Kalsom Yusof
Universiti Sains Malaysia
16 shared publications Hiroyuki Iida
6 shared publications- Naying Gao
School of Information Science, Japan Advanced Institute of Science and Technology, Ishikawa, Japan
5 shared publications - Xiaohan Kang
School of Information Science, Japan Advanced Institute of Science and Technology, Nomi 923-1292, Ishikawa, Japan
4 shared publications - Zeliang Zhang
School of Information Science, Japan Advanced Institute of Science and Technology, Ishikawa, Japan
4 shared publications
Latest Publications
- 1 August 2026
- Published by IOP Publishing in Engineering Research Express
- Vol. 8 (16) , 165222
A coordinated hybrid initialization and multi-operator framework for swarm-based optimization
Population-based metaheuristic algorithms are widely applied to both continuous and discrete optimization problems; however, their performance is often constrained by limitations in population initialization quality and the coordination between exploration and exploitation, particularly in complex and high-dimensional optimization landscapes. In many existing approaches, uniform random initialization leads to uneven spatial coverage in continuous domains and insufficient solution diversity, while single-mechanism search strategies often fail to provide an effective transition between global exploration and local exploitation. These structural limitations reduce search robustness and increase the risk of premature convergence, especially in multimodal, hybrid, and composition optimization problems. To address these challenges, this study proposes a coordinated enhancement framework that jointly improves population initialization and search coordination. The proposed framework integrates a hybrid dual-population initialization strategy with a threshold-based multi-operator search mechanism. During initialization, uniform random sampling and hyperbolic transformation are combined with superiority-based selection to generate a high-quality and diverse initial population. Throughout the optimization process, cosine contraction, sinusoidal perturbation operator, and the original SCSA search operator are coordinated using an iteration-regulated search-strength coefficient with predefined threshold intervals, enabling a structured transition between exploration and exploitation. Based on this framework, an enhanced sand cat swarm algorithm (ESC-SCSA) is developed and comprehensively evaluated through sensitivity analysis, ablation studies, convergence analysis, and statistical significance tests on the CEC2017 and CEC2022 benchmark suites, followed by validation on flexible job shop scheduling problem instances. Experimental results demonstrate that ESC-SCSA consistently improves the original SCSA and exhibits competitive performance compared with several state-of-the-art metaheuristic algorithms across the CEC2017 and CEC2022 benchmark suites. Sensitivity analysis identifies effective parameter settings for the proposed framework, while the ablation study confirms that the hybrid initialization strategy and threshold-based multi-operator coordination each contribute substantially to the observed performance improvements. Convergence analysis further reveals faster and more stable search behaviour throughout the optimization process. Moreover, ESC-SCSA achieved the lowest mean makespan on nine of the ten Brandimarte benchmark instances, indicating its effectiveness and potential generalization capability across both continuous and discrete optimization problems.
- 1 August 2026
- Published by IOP Publishing in Engineering Research Express
- Vol. 8 (16) , 165222
A coordinated hybrid initialization and multi-operator framework for swarm-based optimization
Population-based metaheuristic algorithms are widely applied to both continuous and discrete optimization problems; however, their performance is often constrained by limitations in population initialization quality and the coordination between exploration and exploitation, particularly in complex and high-dimensional optimization landscapes. In many existing approaches, uniform random initialization leads to uneven spatial coverage in continuous domains and insufficient solution diversity, while single-mechanism search strategies often fail to provide an effective transition between global exploration and local exploitation. These structural limitations reduce search robustness and increase the risk of premature convergence, especially in multimodal, hybrid, and composition optimization problems. To address these challenges, this study proposes a coordinated enhancement framework that jointly improves population initialization and search coordination. The proposed framework integrates a hybrid dual-population initialization strategy with a threshold-based multi-operator search mechanism. During initialization, uniform random sampling and hyperbolic transformation are combined with superiority-based selection to generate a high-quality and diverse initial population. Throughout the optimization process, cosine contraction, sinusoidal perturbation operator, and the original SCSA search operator are coordinated using an iteration-regulated search-strength coefficient with predefined threshold intervals, enabling a structured transition between exploration and exploitation. Based on this framework, an enhanced sand cat swarm algorithm (ESC-SCSA) is developed and comprehensively evaluated through sensitivity analysis, ablation studies, convergence analysis, and statistical significance tests on the CEC2017 and CEC2022 benchmark suites, followed by validation on flexible job shop scheduling problem instances. Experimental results demonstrate that ESC-SCSA consistently improves the original SCSA and exhibits competitive performance compared with several state-of-the-art metaheuristic algorithms across the CEC2017 and CEC2022 benchmark suites. Sensitivity analysis identifies effective parameter settings for the proposed framework, while the ablation study confirms that the hybrid initialization strategy and threshold-based multi-operator coordination each contribute substantially to the observed performance improvements. Convergence analysis further reveals faster and more stable search behaviour throughout the optimization process. Moreover, ESC-SCSA achieved the lowest mean makespan on nine of the ten Brandimarte benchmark instances, indicating its effectiveness and potential generalization capability across both continuous and discrete optimization problems.
- Open Access
- 28 November 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 42745
A novel deep neural model for efficient and scalable historical place image classification
Historical place image classification is a crucial intersection of artificial intelligence and cultural heritage preservation, essential for safeguarding architectural legacies in the digital age. Current AI classification systems often struggle with generalizability across regions and styles, require substantial computational resources, and lack interpretability for heritage professionals. This research introduces HistoNet, a hybrid deep learning framework designed to overcome these limitations. HistoNet integrates convolutional neural networks for local feature extraction and Transformer mechanisms for global context, while also using Mamba state-space models, which operate with linear time complexity O(L), thereby offering substantially higher computational efficiency than Transformer architectures that scale quadratically O(\(\hbox {L}^{2}\)). Enhanced with Squeeze-and-Excitation attention blocks and SHapley Additive exPlanations based interpretability, it provides pixel-level attribution maps for more accurate validation of classification decisions. Deployed through a user-friendly web interface with drag-and-drop functionality, HistoNet democratizes access by enabling researchers, conservationists, educators, and heritage professionals in resource-constrained regions to leverage advanced AI tools without requiring high-end hardware or specialized expertise. Testing shows that it achieves 95.66% accuracy on the Architectural Heritage Elements dataset and 96.46% on the Historical Building Recognition dataset, significantly outperforming models such as VGG16 and ResNet50. Moreover, the system demonstrates robust generalization, maintaining stable performance across Malacca’s diverse lighting conditions, weather variations, and architectural styles, thereby ensuring practical reliability in real-world heritage environments. By combining efficiency, accessibility, and robustness, this research makes advanced heritage classification broadly usable and reliable, promoting sustainable development goals in cultural heritage conservation and bridging digital divides.- 28 November 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 42745
Open AccessA novel deep neural model for efficient and scalable historical place image classification
Historical place image classification is a crucial intersection of artificial intelligence and cultural heritage preservation, essential for safeguarding architectural legacies in the digital age. Current AI classification systems often struggle with generalizability across regions and styles, require substantial computational resources, and lack interpretability for heritage professionals. This research introduces HistoNet, a hybrid deep learning framework designed to overcome these limitations. HistoNet integrates convolutional neural networks for local feature extraction and Transformer mechanisms for global context, while also using Mamba state-space models, which operate with linear time complexity O(L), thereby offering substantially higher computational efficiency than Transformer architectures that scale quadratically O(\(\hbox {L}^{2}\)). Enhanced with Squeeze-and-Excitation attention blocks and SHapley Additive exPlanations based interpretability, it provides pixel-level attribution maps for more accurate validation of classification decisions. Deployed through a user-friendly web interface with drag-and-drop functionality, HistoNet democratizes access by enabling researchers, conservationists, educators, and heritage professionals in resource-constrained regions to leverage advanced AI tools without requiring high-end hardware or specialized expertise. Testing shows that it achieves 95.66% accuracy on the Architectural Heritage Elements dataset and 96.46% on the Historical Building Recognition dataset, significantly outperforming models such as VGG16 and ResNet50. Moreover, the system demonstrates robust generalization, maintaining stable performance across Malacca’s diverse lighting conditions, weather variations, and architectural styles, thereby ensuring practical reliability in real-world heritage environments. By combining efficiency, accessibility, and robustness, this research makes advanced heritage classification broadly usable and reliable, promoting sustainable development goals in cultural heritage conservation and bridging digital divides. - Open Access
- 13 October 2025
- Published by MDPI AG in Information
- Vol. 16 (10) , 890
Inherent Addiction Mechanisms in Video Games’ Gacha
Gacha games, particularly those using Free-to-Play (F2P) models, have become increasingly popular yet controversial due to their addictive mechanics, often likened to gambling. This study investigates the inherent addictive mechanisms of Gacha games, focusing on Genshin Impact, a leading title in the genre. We analyze the interplay between reward frequency, game attractiveness, and player addiction using the Game Refinement theory and the Motion in Mind framework. Our analysis identifies a critical threshold at approximately 55 pulls per rare item (), with a corresponding gravity-in-mind value of 7.4. Beyond this point, the system exhibits gambling-like dynamics, as indicated by Game Refinement and Motion in Mind metrics. This threshold was measured using empirical gacha data collected from Genshin Impact players and analyzed through theoretical models. While not claiming direct causal evidence of player behavior change, the results highlight a measurable boundary where structural design risks fostering addiction-like compulsion. The study contributes theoretical insights with ethical implications for game design, by identifying critical thresholds in reward frequency and game dynamics that mark the shift toward gambling-like reinforcement. The methodologies, including quantitative analysis and empirical data, ensure robust results contributing to responsible digital entertainment discourse.- 13 October 2025
- Published by MDPI AG in Information
- Vol. 16 (10) , 890
Open AccessInherent Addiction Mechanisms in Video Games’ Gacha
Gacha games, particularly those using Free-to-Play (F2P) models, have become increasingly popular yet controversial due to their addictive mechanics, often likened to gambling. This study investigates the inherent addictive mechanisms of Gacha games, focusing on Genshin Impact, a leading title in the genre. We analyze the interplay between reward frequency, game attractiveness, and player addiction using the Game Refinement theory and the Motion in Mind framework. Our analysis identifies a critical threshold at approximately 55 pulls per rare item (), with a corresponding gravity-in-mind value of 7.4. Beyond this point, the system exhibits gambling-like dynamics, as indicated by Game Refinement and Motion in Mind metrics. This threshold was measured using empirical gacha data collected from Genshin Impact players and analyzed through theoretical models. While not claiming direct causal evidence of player behavior change, the results highlight a measurable boundary where structural design risks fostering addiction-like compulsion. The study contributes theoretical insights with ethical implications for game design, by identifying critical thresholds in reward frequency and game dynamics that mark the shift toward gambling-like reinforcement. The methodologies, including quantitative analysis and empirical data, ensure robust results contributing to responsible digital entertainment discourse. - Open Access
- 26 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 172326-172340
From Opening to Endgame: When Does the Game Get Engaging and Exciting? Exploring Phase-Based Velocity Dynamics Using the MiM Framework
Understanding how engagement and excitement evolve throughout different phases of a sporting event is essential for advancing research in sports analytics, psychology, and game design. Previous models, like Game Refinement theory, have effectively quantified overall game sophistication but often overlook the dynamic, temporal nature of gameplay. This study addresses this gap by applying the physics-inspired Motion in Mind (MiM) framework to explicitly examine phase-specific gameplay dynamics in three globally popular sports: ODI cricket, T20 cricket, and soccer. Each match was segmented into opening, middle, and endgame phases, enabling detailed computation of key MiM parameters including velocity, acceleration, jerk, momentum, and potential reinforcement energy. Notably, ODI cricket showed initial high excitement that gradually declined toward the end, highlighting strategic shifts from aggressive to more calculated gameplay, whereas T20 cricket exhibited rising momentum and heightened excitement towards the endgame. In contrast, soccer unexpectedly revealed intense peaks during its middle phase. These findings provide novel theoretical insights into how temporal structures shape spectator and player engagement and offer practical strategies for optimizing match design, coaching tactics, and spectator enjoyment across competitive sports contexts.
- 26 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 172326-172340
Open AccessFrom Opening to Endgame: When Does the Game Get Engaging and Exciting? Exploring Phase-Based Velocity Dynamics Using the MiM Framework
Understanding how engagement and excitement evolve throughout different phases of a sporting event is essential for advancing research in sports analytics, psychology, and game design. Previous models, like Game Refinement theory, have effectively quantified overall game sophistication but often overlook the dynamic, temporal nature of gameplay. This study addresses this gap by applying the physics-inspired Motion in Mind (MiM) framework to explicitly examine phase-specific gameplay dynamics in three globally popular sports: ODI cricket, T20 cricket, and soccer. Each match was segmented into opening, middle, and endgame phases, enabling detailed computation of key MiM parameters including velocity, acceleration, jerk, momentum, and potential reinforcement energy. Notably, ODI cricket showed initial high excitement that gradually declined toward the end, highlighting strategic shifts from aggressive to more calculated gameplay, whereas T20 cricket exhibited rising momentum and heightened excitement towards the endgame. In contrast, soccer unexpectedly revealed intense peaks during its middle phase. These findings provide novel theoretical insights into how temporal structures shape spectator and player engagement and offer practical strategies for optimizing match design, coaching tactics, and spectator enjoyment across competitive sports contexts.
- 22 August 2025
- Published by Springer Nature
- p. 468-477
Simulating AI-Human Collaborative Strategies in Sudoku Puzzle Generation
- 22 August 2025
- Published by Springer Nature
- p. 468-477
Simulating AI-Human Collaborative Strategies in Sudoku Puzzle Generation
- Open Access
- 17 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 103264-103281
Modeling Sports Engagement: A Game Refinement Theory Perspective on Game Length and Scoring Frequency
Game structure plays a crucial role in shaping engagement for both players and spectators, with elements such as game length, scoring frequency, and outcome uncertainty significantly influencing the emotional and cognitive experience. This paper investigates how these structural components impact engagement across three globally popular sports, One Day International (ODI) cricket, Twenty20 (T20) cricket, and soccer. Using Game Refinement (GR) theory and the Motion in Mind (MiM) framework, we quantitatively model engagement dynamics by analyzing the game length, scoring patterns, uncertainty resolution, and psychological pacing. Our results reveal that ODI cricket falls within the optimal GR zone (GR ∈ {0.07, 0.08}), indicating a balanced structure that supports fairness and sustained engagement. T20 cricket, while exceeding the optimal GR range, achieves high emotional intensity due to its rapid scoring and momentum shifts. Soccer, in contrast, exhibits a low scoring rate and a GR value above the ideal range, reflecting higher unpredictability and reduced fairness. To optimize soccer’s game length and scoring frequency, we propose a hypothetical 60-minute match model. Our findings demonstrate that reducing game length while scaling scoring frequency in soccer leads to an ideal GR value (≈ 0.077), aligning structural fairness with emotional pacing. This study offers a novel computational and theoretical lens for evaluating and refining sports formats, with broader implications for enhancing engagement across competitive sports.
- 17 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 103264-103281
Open AccessModeling Sports Engagement: A Game Refinement Theory Perspective on Game Length and Scoring Frequency
Game structure plays a crucial role in shaping engagement for both players and spectators, with elements such as game length, scoring frequency, and outcome uncertainty significantly influencing the emotional and cognitive experience. This paper investigates how these structural components impact engagement across three globally popular sports, One Day International (ODI) cricket, Twenty20 (T20) cricket, and soccer. Using Game Refinement (GR) theory and the Motion in Mind (MiM) framework, we quantitatively model engagement dynamics by analyzing the game length, scoring patterns, uncertainty resolution, and psychological pacing. Our results reveal that ODI cricket falls within the optimal GR zone (GR ∈ {0.07, 0.08}), indicating a balanced structure that supports fairness and sustained engagement. T20 cricket, while exceeding the optimal GR range, achieves high emotional intensity due to its rapid scoring and momentum shifts. Soccer, in contrast, exhibits a low scoring rate and a GR value above the ideal range, reflecting higher unpredictability and reduced fairness. To optimize soccer’s game length and scoring frequency, we propose a hypothetical 60-minute match model. Our findings demonstrate that reducing game length while scaling scoring frequency in soccer leads to an ideal GR value (≈ 0.077), aligning structural fairness with emotional pacing. This study offers a novel computational and theoretical lens for evaluating and refining sports formats, with broader implications for enhancing engagement across competitive sports.
- Open Access
- 21 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 90433-90466
Machine Learning and Deep Learning Approaches for Fake News Detection: A Systematic Review of Techniques, Challenges, and Advancements
In response to the escalating threat of fake news on social media, this systematic literature review analyzes the recent advancements in machine learning and deep learning approaches for automated detection. Following the PRISMA guidelines, we examined 90 peer-reviewed studies published between 2020 and 2024 to evaluate the model effectiveness, identify limitations, and highlight emerging trends. Our analysis shows that deep learning models, particularly transformer-based architectures such as BERT, consistently outperform traditional machine learning methods, often achieving a high accuracy (Acc), precision (P), recall (R), and F1-score (F1). For instance, a BERT-based model reported up to 99.9% accuracy on the Kaggle fake news dataset and above 98% accuracy on other public datasets, including ISOT, Fake-or-Real, and D3. Similarly, the GANM model demonstrated robust performance on the FakeNewsNet dataset by integrating text and social features. Transfer learning and multimodal models that incorporate user behaviour and network information significantly improve detection in diverse, low-resource environments. However, challenges persist in terms of the dataset quality, model interpretability, domain generalisability, and real-time deployment. This review also underscores the limited adoption of few-shot and zero-shot learning techniques, highlighting a promising direction for future research on handling emerging misinformation using minimal training data. To support practical deployment, we advocate the development of explainable, multilingual, and lightweight models with greater emphasis on human-centred evaluation and ethical considerations. Our findings provide a foundation for researchers and practitioners to build scalable, trustworthy, and context-aware fake news detection systems for global use.
- 21 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 90433-90466
Open AccessMachine Learning and Deep Learning Approaches for Fake News Detection: A Systematic Review of Techniques, Challenges, and Advancements
In response to the escalating threat of fake news on social media, this systematic literature review analyzes the recent advancements in machine learning and deep learning approaches for automated detection. Following the PRISMA guidelines, we examined 90 peer-reviewed studies published between 2020 and 2024 to evaluate the model effectiveness, identify limitations, and highlight emerging trends. Our analysis shows that deep learning models, particularly transformer-based architectures such as BERT, consistently outperform traditional machine learning methods, often achieving a high accuracy (Acc), precision (P), recall (R), and F1-score (F1). For instance, a BERT-based model reported up to 99.9% accuracy on the Kaggle fake news dataset and above 98% accuracy on other public datasets, including ISOT, Fake-or-Real, and D3. Similarly, the GANM model demonstrated robust performance on the FakeNewsNet dataset by integrating text and social features. Transfer learning and multimodal models that incorporate user behaviour and network information significantly improve detection in diverse, low-resource environments. However, challenges persist in terms of the dataset quality, model interpretability, domain generalisability, and real-time deployment. This review also underscores the limited adoption of few-shot and zero-shot learning techniques, highlighting a promising direction for future research on handling emerging misinformation using minimal training data. To support practical deployment, we advocate the development of explainable, multilingual, and lightweight models with greater emphasis on human-centred evaluation and ethical considerations. Our findings provide a foundation for researchers and practitioners to build scalable, trustworthy, and context-aware fake news detection systems for global use.
- 5 May 2025
- Published by Elsevier in Entertainment Computing
- Vol. 54
Enhancing auction experiences: Game dynamics and customer experience design
- 5 May 2025
- Published by Elsevier in Entertainment Computing
- Vol. 54
Enhancing auction experiences: Game dynamics and customer experience design
- 26 February 2025
- Published by Elsevier in Neurocomputing
- Vol. 632
Words shaping worlds: A comprehensive exploration of text-driven image and video generation with generative adversarial networks
- 26 February 2025
- Published by Elsevier in Neurocomputing
- Vol. 632
Words shaping worlds: A comprehensive exploration of text-driven image and video generation with generative adversarial networks
- 12 November 2024
- Published by Elsevier in Entertainment Computing
- Vol. 52
Ludogenic innovation: How playing incentivizes technological innovation
- 12 November 2024
- Published by Elsevier in Entertainment Computing
- Vol. 52
Ludogenic innovation: How playing incentivizes technological innovation