Malek Barhoush
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Malek Barhoush has a total of 43 co-authors. Here are the top co-authors by number of shared publications:
- Ayoub Alsarhan
Department of Data Science and Artificial Intelligence, Faculty of Information Technology, Al-Ahliyya Amman University, Amman 19111, Jordan
6 shared publications - Mamoon Obiedat
Department of Information Technology, Faculty of Prince Al-Hussein Bin Abdallah II for Information Technology, The Hashemite University, Zarqa 13133, Jordan
4 shared publications - Bilal H. Abed Alguni
Department of Computer Sciences, Yarmouk University, Irbid, Jordan
3 shared publications - Osama Harfoushi
Department of Computer Information Science, Higher Colleges of Technology, Sharjah P.O. Box 25026, United Arab Emirates
3 shared publications - Sami Aziz Alshammari
Department of Information Technology, Faculty of Computing and Information Technology, Northern Border University, Rafha 76313, Saudi Arabia
3 shared publications - Khalid Hamad Alnafisah
Department of Computer Sciences, Faculty of Computing and Information Technology, Northern Border University, Rafha 73213, Saudi Arabia
2 shared publications
Latest Publications
- Open Access
Leakage-Controlled Evaluation of CNN-LSTM Models for Process Temperature Forecasting and Soft Sensing in Machining: Evidence from the AI4I 2020 Benchmark
Convolutional recurrent networks are widely reported as being able to forecast process temperature in machining accurately. Many such evaluations, however, split overlapping sliding windows at random, which lets near-duplicate samples cross the train–test boundary. We re-examine process temperature forecasting and soft sensing on the AI4I 2020 benchmark under a leakage-controlled protocol: a chronological split with a purge gap, training-only scaling, five seeds, block-bootstrap confidence intervals and Diebold–Mariano tests with an autocorrelation-robust variance. LSTM, GRU, CNN-LSTM and an attention-augmented CNN-LSTM are compared with persistence, linear, random forest and multilayer-perceptron baselines that receive the same ten-record window. Under a random split the attention model reaches ; under the chronological split it falls to . Rolling-origin cross-validation, a purge-gap sweep, a factorial ablation and sensitivity analyses support this picture. In soft sensing, the error is dominated by a drifting process–air temperature offset that the machine signals in this benchmark do not observe, and refitting on recent data matters more than architecture. When lagged process temperature is available, persistence (MAE 0.054 K) is not beaten at horizons of 1–30 records, and convolutional and attention stages add no benefit while costing up to 2.4 times the training time of a plain LSTM. On this benchmark, the evaluation protocol changed the conclusions more than did the choice of architecture. Claimed gains of deep sequence models for machine monitoring should therefore be reported against persistence and information-matched baselines under a time-ordered split.Open AccessLeakage-Controlled Evaluation of CNN-LSTM Models for Process Temperature Forecasting and Soft Sensing in Machining: Evidence from the AI4I 2020 Benchmark
Convolutional recurrent networks are widely reported as being able to forecast process temperature in machining accurately. Many such evaluations, however, split overlapping sliding windows at random, which lets near-duplicate samples cross the train–test boundary. We re-examine process temperature forecasting and soft sensing on the AI4I 2020 benchmark under a leakage-controlled protocol: a chronological split with a purge gap, training-only scaling, five seeds, block-bootstrap confidence intervals and Diebold–Mariano tests with an autocorrelation-robust variance. LSTM, GRU, CNN-LSTM and an attention-augmented CNN-LSTM are compared with persistence, linear, random forest and multilayer-perceptron baselines that receive the same ten-record window. Under a random split the attention model reaches ; under the chronological split it falls to . Rolling-origin cross-validation, a purge-gap sweep, a factorial ablation and sensitivity analyses support this picture. In soft sensing, the error is dominated by a drifting process–air temperature offset that the machine signals in this benchmark do not observe, and refitting on recent data matters more than architecture. When lagged process temperature is available, persistence (MAE 0.054 K) is not beaten at horizons of 1–30 records, and convolutional and attention stages add no benefit while costing up to 2.4 times the training time of a plain LSTM. On this benchmark, the evaluation protocol changed the conclusions more than did the choice of architecture. Claimed gains of deep sequence models for machine monitoring should therefore be reported against persistence and information-matched baselines under a time-ordered split. - Open Access
- 16 November 2026
- Published by MDPI AG in Future Internet
- Vol. 18 (11) , 632
Auditing a Conditional-VAE Synthetic-Data Firewall for Privacy-Preserving Deep Learning
Privacy-sensitive machine learning programmes need deployable models without repeatedly circulating original records among developers, analysts and downstream services. This paper evaluates a class-conditional variational autoencoder (CVAE) placed as a synthetic-data firewall between a protected corpus and downstream learners that never see the original records, and contributes an audit and release protocol for such a pipeline. The protocol separates four adversary tiers (model, release, sampling and white-box generator access), calibrates membership attacks with 32 shadow pipelines, adds deliberately vulnerable positive controls, tests copying in the lower tail of the distance distribution, and gates the release on utility, exposure, copying and worst-class retention with an exposure threshold anchored to chance. On the digit corpus, over ten independent partitions, synthetic-only training retains % (CNN) and % (GRU) of real-data accuracy. On digits and a clinical table, likelihood-ratio attacks detect leakage in real-data models but not in release-trained models, and memorising learners are protected on all three corpora (including a face-image set). The firewall relocates rather than removes risk: adversaries holding the release or the generator retain a measurable membership signal, and with very small cohorts, the generator passes it through. Conditional GAN, Gaussian-mixture and differentially private baselines, alternative likelihoods, and an independent selection protocol (% for the selected configuration) complete the evaluation; empirical resistance is not a formal guarantee.- 16 November 2026
- Published by MDPI AG in Future Internet
- Vol. 18 (11) , 632
Open AccessAuditing a Conditional-VAE Synthetic-Data Firewall for Privacy-Preserving Deep Learning
Privacy-sensitive machine learning programmes need deployable models without repeatedly circulating original records among developers, analysts and downstream services. This paper evaluates a class-conditional variational autoencoder (CVAE) placed as a synthetic-data firewall between a protected corpus and downstream learners that never see the original records, and contributes an audit and release protocol for such a pipeline. The protocol separates four adversary tiers (model, release, sampling and white-box generator access), calibrates membership attacks with 32 shadow pipelines, adds deliberately vulnerable positive controls, tests copying in the lower tail of the distance distribution, and gates the release on utility, exposure, copying and worst-class retention with an exposure threshold anchored to chance. On the digit corpus, over ten independent partitions, synthetic-only training retains % (CNN) and % (GRU) of real-data accuracy. On digits and a clinical table, likelihood-ratio attacks detect leakage in real-data models but not in release-trained models, and memorising learners are protected on all three corpora (including a face-image set). The firewall relocates rather than removes risk: adversaries holding the release or the generator retain a measurable membership signal, and with very small cohorts, the generator passes it through. Conditional GAN, Gaussian-mixture and differentially private baselines, alternative likelihoods, and an independent selection protocol (% for the selected configuration) complete the evaluation; empirical resistance is not a formal guarantee. - Open Access
- 3 November 2026
- Published by MDPI AG in Future Internet
- Vol. 18 (11) , 604
Certified Envelope Bounds and an Envelope-Informed Learning Surrogate for Road Network Resilience Under Stochastic Capacity Degradation
Road networks are routinely exposed to partial capacity losses caused by incidents, work zones, flooding and pavement failures, and planners need fast, trustworthy estimates of how such losses propagate through route choice. This paper develops a framework for user-equilibrium (UE) traffic networks with degradable link capacities that combines variational analysis with a learned surrogate. Writing the Beckmann potential in the coordinates makes it affine in u, so standard results (the concavity of a pointwise minimum of affine functions and Danskin’s theorem) show that the optimal potential is concave, non-decreasing and differentiable, with a gradient available from one equilibrium solution. The resulting criticality index equals times the congestion delay of each link; the theory adds to this classical indicator a derivative interpretation and certified bounds. We derive two-sided certificates on the potential change for any degradation, the identity , a solver-free certificate for the total system travel time (TSTT), and an additive certificate for restoring sets of links, whose per-link weight is the damaged-state delay of the link times a severity factor, and we use these quantities as features of an envelope-informed surrogate (EIGS), a gradient-boosted model on per-link envelope features diffused over the line graph. Experiments on the Sioux Falls and Anaheim benchmarks show where the results are useful and where they are not. The potential certificate is within a factor of 1.45 of the exact change at 25% capacity loss on Sioux Falls (1.03 on the 100 most critical Anaheim links) but loosens quickly with severity (factor 3.28 at 50% loss on Sioux Falls; 574 and 74 at 90% loss), and the solver-free TSTT bound exceeds the true change by a median factor of 3.8–12 in the multi-link Monte Carlo scenario sets. No single cheap indicator ranks link criticality best on both networks. On Sioux Falls, EIGS predicts the TSTT increase with (0.913–0.960 over ten splits) against 0.919 for ridge regression and 0.924 for gradient boosting on the diffused severities, and with against 0.862 on a spatially correlated test set generated after the model was fixed. Under shifts in demand () and nominal capacity, refreshing its envelope features from one equilibrium of the shifted network gives –; using the same equilibrium to rescale the predictions of raw-input baselines gives them –, so the gain from the per-link envelope features is modest. On Anaheim, whose losses are extremely heavy-tailed, EIGS ranks scenarios well (Spearman ) but does not predict their magnitude reliably, while ridge regression with a cross-validated penalty ranks them poorly (). For restoring three of about fifteen damaged links, the certified envelope rule, which needs a single equilibrium, recovers on average 93.6% of the loss recovered by the exact optimum over all three-link sets.- 3 November 2026
- Published by MDPI AG in Future Internet
- Vol. 18 (11) , 604
Open AccessCertified Envelope Bounds and an Envelope-Informed Learning Surrogate for Road Network Resilience Under Stochastic Capacity Degradation
Road networks are routinely exposed to partial capacity losses caused by incidents, work zones, flooding and pavement failures, and planners need fast, trustworthy estimates of how such losses propagate through route choice. This paper develops a framework for user-equilibrium (UE) traffic networks with degradable link capacities that combines variational analysis with a learned surrogate. Writing the Beckmann potential in the coordinates makes it affine in u, so standard results (the concavity of a pointwise minimum of affine functions and Danskin’s theorem) show that the optimal potential is concave, non-decreasing and differentiable, with a gradient available from one equilibrium solution. The resulting criticality index equals times the congestion delay of each link; the theory adds to this classical indicator a derivative interpretation and certified bounds. We derive two-sided certificates on the potential change for any degradation, the identity , a solver-free certificate for the total system travel time (TSTT), and an additive certificate for restoring sets of links, whose per-link weight is the damaged-state delay of the link times a severity factor, and we use these quantities as features of an envelope-informed surrogate (EIGS), a gradient-boosted model on per-link envelope features diffused over the line graph. Experiments on the Sioux Falls and Anaheim benchmarks show where the results are useful and where they are not. The potential certificate is within a factor of 1.45 of the exact change at 25% capacity loss on Sioux Falls (1.03 on the 100 most critical Anaheim links) but loosens quickly with severity (factor 3.28 at 50% loss on Sioux Falls; 574 and 74 at 90% loss), and the solver-free TSTT bound exceeds the true change by a median factor of 3.8–12 in the multi-link Monte Carlo scenario sets. No single cheap indicator ranks link criticality best on both networks. On Sioux Falls, EIGS predicts the TSTT increase with (0.913–0.960 over ten splits) against 0.919 for ridge regression and 0.924 for gradient boosting on the diffused severities, and with against 0.862 on a spatially correlated test set generated after the model was fixed. Under shifts in demand () and nominal capacity, refreshing its envelope features from one equilibrium of the shifted network gives –; using the same equilibrium to rescale the predictions of raw-input baselines gives them –, so the gain from the per-link envelope features is modest. On Anaheim, whose losses are extremely heavy-tailed, EIGS ranks scenarios well (Spearman ) but does not predict their magnitude reliably, while ridge regression with a cross-validated penalty ranks them poorly (). For restoring three of about fifteen damaged links, the certified envelope rule, which needs a single equilibrium, recovers on average 93.6% of the loss recovered by the exact optimum over all three-link sets. - Open Access
A Novel Distributed Smart Water IoT-SCADA Simulation Testbed for Graph-Temporal AI Detection of Cross-Zone Process–Perception Dissociation Attacks
Smart water infrastructure increasingly relies on distributed Industrial Internet of Things (IIoT) and SCADA architectures for real-time monitoring, supervisory control, and operational decision-making. However, this integration also creates cyber-physical vulnerabilities in which attackers can maintain plausible communication and supervisory feedback while the underlying process gradually deviates from its true state. This paper presents a distributed smart water IoT-SCADA simulation testbed for investigating Cross-Zone Process–Perception Dissociation Attacks (CZPPDAs), where apparently normal SCADA/HMI information conceals harmful physical-process drift. The testbed spans external threat, perimeter, DMZ, enterprise IT, supervisory control, OT control, field process, and monitoring/data-acquisition layers, supporting realistic attacker progression, telemetry interception, semantic manipulation, supervisory deception, and concealed process deviation. A hybrid cyber-physical dataset is generated from communication descriptors, protocol variables, supervisory states, controller attributes, process measurements, and cross-zone consistency indicators. For detection, CZSD-Net models SCADA, HMI, PLC, RTU, and gateway components as an industrial dependency graph and learns evolving process–perception inconsistency through relation-aware graph encoding, dissociation reasoning, and temporal fusion. Experimental evaluation shows progressive improvement over baseline graph-learning configurations. The final model achieves 97.94% testing accuracy, 97.62% precision, 97.29% recall, and 97.43% F1-score. These results show that the proposed testbed and graph-temporal framework provide an effective approach for studying and detecting stealthy cyber-physical deception in smart water IoT-SCADA environments. It also supports reproducible evaluation of detection behavior across synchronized supervisory, control, network, and physical-process observations.
Open AccessA Novel Distributed Smart Water IoT-SCADA Simulation Testbed for Graph-Temporal AI Detection of Cross-Zone Process–Perception Dissociation Attacks
Smart water infrastructure increasingly relies on distributed Industrial Internet of Things (IIoT) and SCADA architectures for real-time monitoring, supervisory control, and operational decision-making. However, this integration also creates cyber-physical vulnerabilities in which attackers can maintain plausible communication and supervisory feedback while the underlying process gradually deviates from its true state. This paper presents a distributed smart water IoT-SCADA simulation testbed for investigating Cross-Zone Process–Perception Dissociation Attacks (CZPPDAs), where apparently normal SCADA/HMI information conceals harmful physical-process drift. The testbed spans external threat, perimeter, DMZ, enterprise IT, supervisory control, OT control, field process, and monitoring/data-acquisition layers, supporting realistic attacker progression, telemetry interception, semantic manipulation, supervisory deception, and concealed process deviation. A hybrid cyber-physical dataset is generated from communication descriptors, protocol variables, supervisory states, controller attributes, process measurements, and cross-zone consistency indicators. For detection, CZSD-Net models SCADA, HMI, PLC, RTU, and gateway components as an industrial dependency graph and learns evolving process–perception inconsistency through relation-aware graph encoding, dissociation reasoning, and temporal fusion. Experimental evaluation shows progressive improvement over baseline graph-learning configurations. The final model achieves 97.94% testing accuracy, 97.62% precision, 97.29% recall, and 97.43% F1-score. These results show that the proposed testbed and graph-temporal framework provide an effective approach for studying and detecting stealthy cyber-physical deception in smart water IoT-SCADA environments. It also supports reproducible evaluation of detection behavior across synchronized supervisory, control, network, and physical-process observations.
- Open Access
Renewable Energy-Aware Carbon Digital Twin for Smart and Sustainable Construction Operations Using AI-Based Multi-Objective Optimization
Smart and sustainable construction requires methods that jointly manage construction processes, carbon emissions, energy supply, and decision-making. This study develops a renewable energy (RE)-aware carbon digital twin using Siemens Tecnomatix Plant Simulation, version 2022 (Siemens Digital Industries Software, Plano, TX, USA), hereafter referred to as STPS, and artificial intelligence (AI)-based multi-objective optimization. The framework integrates activity records, equipment-demand profiles, renewable-generation data, battery-storage states, emission factors, cost parameters, schedule indicators, and reinforcement learning (RL) state–action–reward records within a discrete-event simulation environment. Construction activities are represented through event-driven source, queue, buffer, processor, resource-pool, transporter, event-controller, table-file, and sink objects, enabling predecessor validation, resource allocation, processing, completion tracking, and key performance indicator (KPI) updates. The energy layer coordinates equipment demand, solar photovoltaic (PV) generation, battery charging and discharging, grid electricity, diesel backup, RE share, battery state of charge (SOC), and emissions. The AI-control layer observes carbon, cost, delay, queue length, utilization, idle time, renewable share, and SOC; filters infeasible actions; evaluates corrective interventions; and ranks policies under carbon, delay, and SOC constraints. Three scenarios are evaluated: diesel–grid baseline operation, rule-based solar-battery operation, and AI-controlled renewable-aware operation. The AI-controlled scenario achieved 16,850 kg carbon dioxide equivalent (kg CO2e), a cost of 2.28 million United States dollars (M USD), a delay of 4.2 h, an 81.6% RE share, 86.7% resource utilization, and 8.6% idle time. Relative to the diesel–grid baseline, it reduced emissions by 32.2%, cost by 18.0%, delay by 66.7%, and diesel-equivalent energy by 97.7%, while increasing RE share by 69.3 percentage points. Reward-weight sensitivity results show that the balanced policy preserves low-carbon performance across stakeholder priorities. The findings demonstrate that integrating digital twins, RE dispatch, and AI-based decision control can support data-driven, optimized low-carbon construction management.
Open AccessRenewable Energy-Aware Carbon Digital Twin for Smart and Sustainable Construction Operations Using AI-Based Multi-Objective Optimization
Smart and sustainable construction requires methods that jointly manage construction processes, carbon emissions, energy supply, and decision-making. This study develops a renewable energy (RE)-aware carbon digital twin using Siemens Tecnomatix Plant Simulation, version 2022 (Siemens Digital Industries Software, Plano, TX, USA), hereafter referred to as STPS, and artificial intelligence (AI)-based multi-objective optimization. The framework integrates activity records, equipment-demand profiles, renewable-generation data, battery-storage states, emission factors, cost parameters, schedule indicators, and reinforcement learning (RL) state–action–reward records within a discrete-event simulation environment. Construction activities are represented through event-driven source, queue, buffer, processor, resource-pool, transporter, event-controller, table-file, and sink objects, enabling predecessor validation, resource allocation, processing, completion tracking, and key performance indicator (KPI) updates. The energy layer coordinates equipment demand, solar photovoltaic (PV) generation, battery charging and discharging, grid electricity, diesel backup, RE share, battery state of charge (SOC), and emissions. The AI-control layer observes carbon, cost, delay, queue length, utilization, idle time, renewable share, and SOC; filters infeasible actions; evaluates corrective interventions; and ranks policies under carbon, delay, and SOC constraints. Three scenarios are evaluated: diesel–grid baseline operation, rule-based solar-battery operation, and AI-controlled renewable-aware operation. The AI-controlled scenario achieved 16,850 kg carbon dioxide equivalent (kg CO2e), a cost of 2.28 million United States dollars (M USD), a delay of 4.2 h, an 81.6% RE share, 86.7% resource utilization, and 8.6% idle time. Relative to the diesel–grid baseline, it reduced emissions by 32.2%, cost by 18.0%, delay by 66.7%, and diesel-equivalent energy by 97.7%, while increasing RE share by 69.3 percentage points. Reward-weight sensitivity results show that the balanced policy preserves low-carbon performance across stakeholder priorities. The findings demonstrate that integrating digital twins, RE dispatch, and AI-based decision control can support data-driven, optimized low-carbon construction management.
- Open Access
- 6 June 2026
- Published by Engineering, Technology & Applied Science Research in Engineering, Technology & Applied Science Research
- Vol. 16 (3) , 35085-35093
Taxonomy and Simulation of Bio-Analogous Multi-Stage Cyberattacks
Today, bio-inspired approaches in cybersecurity represent an active research area in the literature, as these approaches offer adaptability, learning, and robustness to face modern, evolving cyber threats. Traditional Intrusion Detection Systems (IDSs) are increasingly proficient at identifying isolated malicious events, yet they remain vulnerable to coordinated, multi-stage deception that mimics biological social structures. The primary gap in current literature is a lack of formal frameworks that map complex predator-prey dynamics, such as distraction-based deception and fission–fusion coordination, to modern network attack vectors. To address this, we propose a novel taxonomy and realistic discrete-event simulation framework for bio-analogous cyberattacks based on two natural archetypes: the Crow, representing sophisticated deception and decoy-based exfiltration, and the Wild Dog, representing decentralized coordination and sequential hunting. Leveraging the OMNeT++/INET framework and an enterprise-inspired network topology, we emulate how biological perseverance and role-division can be mapped to advanced multi-vector intrusions.
- 6 June 2026
- Published by Engineering, Technology & Applied Science Research in Engineering, Technology & Applied Science Research
- Vol. 16 (3) , 35085-35093
Open AccessTaxonomy and Simulation of Bio-Analogous Multi-Stage Cyberattacks
Today, bio-inspired approaches in cybersecurity represent an active research area in the literature, as these approaches offer adaptability, learning, and robustness to face modern, evolving cyber threats. Traditional Intrusion Detection Systems (IDSs) are increasingly proficient at identifying isolated malicious events, yet they remain vulnerable to coordinated, multi-stage deception that mimics biological social structures. The primary gap in current literature is a lack of formal frameworks that map complex predator-prey dynamics, such as distraction-based deception and fission–fusion coordination, to modern network attack vectors. To address this, we propose a novel taxonomy and realistic discrete-event simulation framework for bio-analogous cyberattacks based on two natural archetypes: the Crow, representing sophisticated deception and decoy-based exfiltration, and the Wild Dog, representing decentralized coordination and sequential hunting. Leveraging the OMNeT++/INET framework and an enterprise-inspired network topology, we emulate how biological perseverance and role-division can be mapped to advanced multi-vector intrusions.
- Open Access
- 4 September 2025
- Published by Elsevier in Egyptian Informatics Journal
- Vol. 31
Traffic sign detection and recognition in Jordan based on machine learning and deep learning
- 4 September 2025
- Published by Elsevier in Egyptian Informatics Journal
- Vol. 31
Open AccessTraffic sign detection and recognition in Jordan based on machine learning and deep learning
- Open Access
- 30 December 2024
- Published by MDPI AG in Diagnostics
- Vol. 15 (1) , 66
Using Machine Learning to Diagnose Autism Based on Eye Tracking Technology
Background/Objectives: One of the key challenges in autism is early diagnosis. Early diagnosis leads to early interventions that improve the condition and not worsen autism in the future. Currently, autism diagnoses are based on monitoring by a doctor or specialist after the child reaches a certain age exceeding three years after the parents observe the child’s abnormal behavior. Methods: The paper aims to find another way to diagnose autism that is effective and earlier than traditional methods of diagnosis. Therefore, we used the Eye Gaze fixes map dataset and Eye Tracking Scanpath dataset (ETSDS) to diagnose Autistic Spectrum Disorder (ASDs), while a subset of the ETSDS was used to recognize autism scores. Results: The experimental results showed that the higher accuracy rate reached 96.1% and 98.0% for the hybrid model on Eye Gaze fixes map datasets and ETSDS, respectively. A higher accuracy rate was reached (98.1%) on the ETSDS used to recognize autism scores. Furthermore, the results showed the outperformer for the proposed method results compared to previous works. Conclusions: This confirms the effectiveness of using artificial intelligence techniques in diagnosing diseases in general and diagnosing autism, in addition to the need to increase research in the field of diagnosing diseases using advanced techniques.
- 30 December 2024
- Published by MDPI AG in Diagnostics
- Vol. 15 (1) , 66
Open AccessUsing Machine Learning to Diagnose Autism Based on Eye Tracking Technology
Background/Objectives: One of the key challenges in autism is early diagnosis. Early diagnosis leads to early interventions that improve the condition and not worsen autism in the future. Currently, autism diagnoses are based on monitoring by a doctor or specialist after the child reaches a certain age exceeding three years after the parents observe the child’s abnormal behavior. Methods: The paper aims to find another way to diagnose autism that is effective and earlier than traditional methods of diagnosis. Therefore, we used the Eye Gaze fixes map dataset and Eye Tracking Scanpath dataset (ETSDS) to diagnose Autistic Spectrum Disorder (ASDs), while a subset of the ETSDS was used to recognize autism scores. Results: The experimental results showed that the higher accuracy rate reached 96.1% and 98.0% for the hybrid model on Eye Gaze fixes map datasets and ETSDS, respectively. A higher accuracy rate was reached (98.1%) on the ETSDS used to recognize autism scores. Furthermore, the results showed the outperformer for the proposed method results compared to previous works. Conclusions: This confirms the effectiveness of using artificial intelligence techniques in diagnosing diseases in general and diagnosing autism, in addition to the need to increase research in the field of diagnosing diseases using advanced techniques.
- Open Access
- 30 September 2024
- Published by MDPI AG in Technologies
- Vol. 12 (10) , 181
Optimizing Cyber Threat Detection in IoT: A Study of Artificial Bee Colony (ABC)-Based Hyperparameter Tuning for Machine Learning
In the rapidly evolving landscape of the Internet of Things (IoT), cybersecurity remains a critical challenge due to the diverse and complex nature of network traffic and the increasing sophistication of cyber threats. This study investigates the application of the Artificial Bee Colony (ABC) algorithm for hyperparameter optimization (HPO) in machine learning classifiers, specifically focusing on Decision Trees, Support Vector Machines (SVM), and K-Nearest Neighbors (KNN) for IoT network traffic analysis and malware detection. Initially, the basic machine learning models demonstrated accuracies ranging from 69.68% to 99.07%, reflecting their limitations in fully adapting to the varied IoT environments. Through the employment of the ABC algorithm for HPO, significant improvements were achieved, with optimized classifiers reaching up to 100% accuracy, precision, recall, and F1-scores in both training and testing stages. These results highlight the profound impact of HPO in refining model decision boundaries, reducing overfitting, and enhancing generalization capabilities, thereby contributing to the development of more robust and adaptive security frameworks for IoT environments. This study further demonstrates the ABC algorithm’s generalizability across different IoT networks and threats, positioning it as a valuable tool for advancing cybersecurity in increasingly complex IoT ecosystems.
- 30 September 2024
- Published by MDPI AG in Technologies
- Vol. 12 (10) , 181
Open AccessOptimizing Cyber Threat Detection in IoT: A Study of Artificial Bee Colony (ABC)-Based Hyperparameter Tuning for Machine Learning
In the rapidly evolving landscape of the Internet of Things (IoT), cybersecurity remains a critical challenge due to the diverse and complex nature of network traffic and the increasing sophistication of cyber threats. This study investigates the application of the Artificial Bee Colony (ABC) algorithm for hyperparameter optimization (HPO) in machine learning classifiers, specifically focusing on Decision Trees, Support Vector Machines (SVM), and K-Nearest Neighbors (KNN) for IoT network traffic analysis and malware detection. Initially, the basic machine learning models demonstrated accuracies ranging from 69.68% to 99.07%, reflecting their limitations in fully adapting to the varied IoT environments. Through the employment of the ABC algorithm for HPO, significant improvements were achieved, with optimized classifiers reaching up to 100% accuracy, precision, recall, and F1-scores in both training and testing stages. These results highlight the profound impact of HPO in refining model decision boundaries, reducing overfitting, and enhancing generalization capabilities, thereby contributing to the development of more robust and adaptive security frameworks for IoT environments. This study further demonstrates the ABC algorithm’s generalizability across different IoT networks and threats, positioning it as a valuable tool for advancing cybersecurity in increasingly complex IoT ecosystems.
- 10 August 2024
- Published by Springer Nature in International Journal of Machine Learning and Cybernetics
- Vol. 15 (12) , 6107-6148
Bi-objective feature selection in high-dimensional datasets using improved binary chimp optimization algorithm
- 10 August 2024
- Published by Springer Nature in International Journal of Machine Learning and Cybernetics
- Vol. 15 (12) , 6107-6148
Bi-objective feature selection in high-dimensional datasets using improved binary chimp optimization algorithm