{"title":"The Paradigm of Hallucinations in AI-driven cybersecurity systems: Understanding taxonomy, classification outcomes, and mitigations","authors":"Aditya K Sood , Sherali Zeadally , EenKee Hong","doi":"10.1016/j.compeleceng.2025.110307","DOIUrl":"10.1016/j.compeleceng.2025.110307","url":null,"abstract":"<div><div>The adoption of AI to solve cybersecurity problems is occurring exponentially. However, AI-driven cybersecurity systems face significant challenges due to the impact of hallucinations in Large Language Models (LLMs). In AI-driven cybersecurity systems, hallucinations refer to instances when an AI model generates fabricated, inaccurate, and misleading information that impacts the security posture of organizations. This failure to recognize and misreport security threats identifies benign activities as malicious, invents insights not grounded to actual cyber threats, and causes real threats to go undetected due to erroneous interpretations. Hallucinations are a critical problem in AI-driven cybersecurity because they can lead to severe vulnerabilities, erode trust in automated systems, and divert resources to address non-existent threats. In cybersecurity, where real-time, accurate insights are vital, hallucinated outputs—such as mistakenly generated alerts, can cause a misallocation of time and resources. It is crucial to address hallucinations by improving LLM accuracy, grounding outputs in real-time data, and implementing human oversight mechanisms to ensure that AI-based cybersecurity systems remain trustworthy, reliable, and capable of defending against sophisticated threats. We present a taxonomy of hallucinations in LLMs for cybersecurity, including mapping LLM responses to classification outcomes (confusion matrix components). Finally, we discuss mitigation strategies to combat hallucinations.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"124 ","pages":"Article 110307"},"PeriodicalIF":4.0,"publicationDate":"2025-04-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143768952","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Fused multi-level attention features with a constraint fusion network for colorectal tissue classification using histopathological images","authors":"Rashi Chauhan , Karnati Mohan , Pradeep Singh","doi":"10.1016/j.compeleceng.2025.110296","DOIUrl":"10.1016/j.compeleceng.2025.110296","url":null,"abstract":"<div><div>Colorectal cancer (CRC) has a significant mortality rate and continues to damage human lives around the world. Early diagnosis/identification of CRC helps to lengthen human life and detour infection. Histopathological examination is a standard method for diagnosing and detecting CRC. Visual inspection of histopathological diagnoses takes longer, and the outcome is based on physicians’ subjective perceptions. Current approaches mainly rely on various varieties of texture characteristics. Moreover, deep learning methodologies’ categorization efficacy surpasses that of medical practitioners. Nevertheless, elevated complexity would diminish the practicality of these strategies. This study employs a unique method known as fused multi-level attention features with a constraint fusion network (FMANet) for accurately classifying colorectal tissue using histopathological images. FMANet comprises modified dilated residual inception feature extraction, channel-wise, and spatial attention components. These components assist in extracting the most pertinent and prejudiced characteristics from the low-level (LL) and high-level (HL) features in the HIs. The constraint fusion technique is employed to weigh and fuse LL and HL attributes adaptively. The proposed model is evaluated by the number of parameters, accuracy, precision, recall, F1 score, and area under the curve. Moreover, the proposed FMANet attained an accuracy of 95% on the colorectal histology dataset and 97% on National Center for Tumor-CRC-hematoxylin & eosin-100K. The proposed method outperforms ten state-of-the-art algorithms on two datasets, as a result, it is believed that this strategy has great potential to aid physicians in clinical diagnoses and reduce the number of experiments by using computer-aided diagnosis to classify CRC cells. The complete system’s source code will be accessible at <span><span>https://github.com/KarnatiMOHAN/FMANet_colorectal-tissue_classification</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"124 ","pages":"Article 110296"},"PeriodicalIF":4.0,"publicationDate":"2025-04-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143768951","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Ensemble learning unlocking point load forecasting accuracy: A novel framework based on two-stage data preprocessing and improved multi-objective optimisation strategy","authors":"Jingmin Luan , Qiyang Li , Yuyan Qiu , Weican Liu","doi":"10.1016/j.compeleceng.2025.110282","DOIUrl":"10.1016/j.compeleceng.2025.110282","url":null,"abstract":"<div><div>Accurate point power forecasts are critical for maintaining the security and stability of the grid. However, Load data is volatile and difficult to predict with high accuracy. To improve the stability and accuracy of the model, we proposed a novel hybrid load forecasting framework consisting of three modules. Firstly, we used <em>Successive Variational Mode Decomposition (SVMD)</em> in the data preprocessing module to extract trend features and denoise the data. To reduce the negative impact of feature redundancy on model predictions, we employed feature selection to identify four essential features to aid model training. Secondly, in the ensemble learning module, we address the limitations of single predictive models by combining the models <em>Back Propagation (BP)</em>, <em>Temporal Convolutional Network (TCN)</em>, <em>Bidirectional Long Short-Term Memory (BiLSTM)</em>, <em>Bidirectional Gated Recurrent Unit (BiGRU)</em> and <em>Transformer</em>. Finally, in the improved multi-objective optimisation algorithm module, we implemented various strategies to enhance the optimisation algorithm. We designed some experiments using three load datasets from Australia. The results demonstrated that the mean absolute percentage error values are below <strong><em>1.25%</em></strong>, with the best value reaching <strong><em>0.9256%</em></strong>. In contrast, the best result for the mean absolute percentage error of the baseline models was <strong><em>1.4431%</em></strong> in New South Wales. This represents a <strong><em>35.9%</em></strong> improvement in load forecasting performance with our proposed model, highlighting its superior accuracy compared to competing approaches. It shows that our forecasting framework is far better than other rivals.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"124 ","pages":"Article 110282"},"PeriodicalIF":4.0,"publicationDate":"2025-04-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143760338","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Secure lightweight digital twin (DT) technology for seamless wireless communication in vehicular ad hoc network","authors":"M.K. Kishore , V. Gajendra Kumar , B. Nancharaiah","doi":"10.1016/j.compeleceng.2025.110291","DOIUrl":"10.1016/j.compeleceng.2025.110291","url":null,"abstract":"<div><div>Security in VANETs (Vehicular Ad Hoc Networks) is crucial due to the inherent vulnerabilities in wireless communication. These networks face several security challenges, including authentication, privacy protection, and secure data transmission. Authentication issues arise from the dynamic nature of vehicular networks, where vehicles need to authenticate each other and entities like roadside units securely. The challenges include GPS spoofing, where attackers manipulate location data to mislead vehicles; Sybil attacks, where a single node illegitimately claims multiple identities to disrupt network operations; and Denial of Service (DoS) attacks, which overwhelm the network, rendering it unavailable for legitimate users. Additionally, key management for secure encryption, replay attacks where previously transmitted messages are maliciously reused, and jamming attacks that disrupt wireless signals are notable threats. To overcome the issue in VANET wireless communication this paper proposed DT-LHBC (Digital Twin Lightweight Hashing Blockchain Cryptography) into VANETs (Vehicular Ad-Hoc Networks) to enhance security and efficiency in vehicular communication systems. DT-LHBC combines advanced cryptographic techniques with digital twin technology to ensure secure data transmission and communication integrity across dynamic vehicular environments. With the implementation of a digital twin technology cryptography model of hashing-based blockchain (LHBC) is integrated to achieve effective wireless communication. Results demonstrate DT-LHBC's effectiveness in mitigating various security threats, including Sybil attacks, Denial of Service (DoS) attacks, and Message Tampering, achieving up to 98 % accuracy in attack detection. Despite challenges in optimizing cryptographic speeds and addressing GPS Spoofing vulnerabilities, DT-LHBC reduces blockchain latency to as low as 12 milliseconds and maintains a high transaction throughput of 500 transactions per second.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"123 ","pages":"Article 110291"},"PeriodicalIF":4.0,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143738038","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"HardSecUAV: A hardware-based mutual authentication protocol for network of drones","authors":"Anubhav Elhence , Vinay Chamola","doi":"10.1016/j.compeleceng.2025.110286","DOIUrl":"10.1016/j.compeleceng.2025.110286","url":null,"abstract":"<div><div>Unmanned Aerial Vehicle (UAV) networks are increasingly utilized in various applications, yet they face significant security challenges due to their open operational environments and resource constraints. Existing authentication protocols often rely on stored secret keys, making them vulnerable to physical attacks and key compromise. To address this research gap, we propose <em>HardSecUAV</em>, a lightweight and secure authentication protocol that leverages hardware-based security features of the DS28S60 cryptographic coprocessor, including Physically Unclonable Functions (PUFs). Our protocol eliminates the need for secret key storage by generating device-specific secrets on-demand, enhancing resistance to physical attacks. Implementation results demonstrate that <em>HardSecUAV</em> achieves a total computation time of approximately <strong>28.775</strong> <span><math><mrow><mi>μ</mi><mi>s</mi></mrow></math></span> during the authentication phase, suitable for real-time operations. The storage requirement is <strong>1024 bits</strong>, and the communication overhead is <strong>1664 bits</strong>, comparable to existing schemes. Compared to existing protocols, <em>HardSecUAV</em> provides enhanced security features, including mutual authentication, replay attack resistance, forward secrecy, and UAV-to-UAV authentication without dependence on Ground Station(GS), making it a robust solution for securing UAV networks.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"123 ","pages":"Article 110286"},"PeriodicalIF":4.0,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143738042","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Cross-domain recommender systems via multimodal domain adaptation","authors":"Adamya Shyam , Ramya Kamani , Venkateswara Rao Kagita , Vikas Kumar","doi":"10.1016/j.compeleceng.2025.110300","DOIUrl":"10.1016/j.compeleceng.2025.110300","url":null,"abstract":"<div><div>Collaborative Filtering (CF) has emerged as one of the most prominent implementation strategies for building recommender systems. The key idea is to exploit the usage patterns of individuals to generate personalized recommendations. CF techniques, especially for newly launched platforms, often face a critical issue known as the data sparsity problem, which greatly limits their performance. Cross-domain CF alleviates the problem of data sparsity by finding a common set of entities (users or items) across the domains, which then act as a conduit for knowledge transfer. Nevertheless, most real-world datasets are collected from different domains, so they often lack information about anchor points or reference information for entity alignment. This paper introduces a domain adaptation technique to align the embeddings of entities across domains. Our approach first exploits the available textual and visual information to independently learn a multi-view latent representation for each entity in the auxiliary and target domains. The different representations of the entity are then fused to generate the corresponding unified representation. A domain classifier is then trained to learn the embedding for the domain alignment by fixing the unified features as the anchor points. Experiments on four publicly available benchmark datasets indicate the effectiveness of our proposed approach.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"123 ","pages":"Article 110300"},"PeriodicalIF":4.0,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143738041","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Diego Feroldi , Pablo Rullo , Sair Rodríguez del Portal, Lautaro Braccia , Patricio Luppi , David Zumoffen
{"title":"Data-driven recursive multivariable modeling, operation, and control of active distribution networks with distributed generation","authors":"Diego Feroldi , Pablo Rullo , Sair Rodríguez del Portal, Lautaro Braccia , Patricio Luppi , David Zumoffen","doi":"10.1016/j.compeleceng.2025.110241","DOIUrl":"10.1016/j.compeleceng.2025.110241","url":null,"abstract":"<div><div>This paper addresses the challenges arising from the increasing integration of distributed generation into active distribution networks (ADNs), focusing on their modeling, operation, and control. Data-driven recursive multivariable modeling, capable of capturing both static and dynamic interactions in real time, has emerged as a promising solution. By utilizing the extensive data generated by modern grid infrastructure, this approach enhances network model accuracy and improves operational efficiency and control strategies. This paper strengthens the connection between Process Systems Engineering (PSE) and power systems, traditionally underexplored in this domain. By integrating PSE principles, particularly data-driven and control allocation methodologies, into the modeling, operation, and control of ADNs, this work optimizes power system performance. Three Recursive Partial Least Squares (RPLS) methodologies—sample-wise, block-wise, and moving-window—are rigorously compared regarding estimation/prediction characteristics and convergence speed. This novel analysis challenges the assumption of instantaneous model adaptation, emphasizing the importance of carefully considering convergence periods for effective monitoring, control, and optimization. The paper proposes and analyzes three control structures integrated into an RPLS-based supervisory strategy for voltage regulation at ADN nodes: (1) decentralized control, (2) control allocation with measurement combination, and (3) optimization-based centralized control. Different integration formats are evaluated based on the controller technology used: (a) simple setpoint updates, (b) full ADN model adaptation to recalculate controller matrices, and (c) full model adaptation for updating the optimization formulation. Simulation results were obtained using the IEEE 33-bus test system. The results reveal a trade-off between the complexity and performance benefits of each control strategy. Although no strategy proves definitively superior, the latter two show more promising overall prospects.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"123 ","pages":"Article 110241"},"PeriodicalIF":4.0,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143738039","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"PrGChain: A privacy-preserving blockchain-enabled energy trading system","authors":"Ahmed-Sami Berkani , Hamouma Moumen , Saber Benharzallah , Mohand Tahar Kechadi , Ahcène Bounceur","doi":"10.1016/j.compeleceng.2025.110289","DOIUrl":"10.1016/j.compeleceng.2025.110289","url":null,"abstract":"<div><div>The integration of blockchain, Internet of Things devices, and distributed energy resources is revolutionizing peer-to-peer energy trading by enabling decentralized, efficient, and transparent transactions. However, existing solutions face challenges related to privacy, interoperability, and scalability. This paper presents PrGChain, a privacy-preserving blockchain-enabled energy trading framework within the smart grid, incorporating a decentralized ZKOracle to securely connect blockchain networks with off-chain energy data sources.</div><div>The proposed ZKOracle employs zero-knowledge proofs to verify energy data without exposing sensitive information, ensuring compliance with privacy regulations, while leveraging a distributed network of oracle nodes for enhanced reliability and interoperability. To improve security and efficiency, PrGChain utilizes smart contracts, decentralized applications (dApps), and leverages stablecoins to mitigate cryptocurrency volatility.</div><div>Performance evaluations demonstrate that our system achieves improved decentralization and privacy without sacrificing efficiency. This is particularly true when deployed on Layer 2 blockchain networks like Polygon, where transaction latency and costs are significantly reduced.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"123 ","pages":"Article 110289"},"PeriodicalIF":4.0,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143737773","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Jie Ma , Hong Chen , Yongming Li , Pin Wang , Chengyu Liu , Yinghua Shen , Witold Pedrycz , Wei Wang , Fan Li
{"title":"Hierarchical manifold sample envelope transformation model for ensemble classification","authors":"Jie Ma , Hong Chen , Yongming Li , Pin Wang , Chengyu Liu , Yinghua Shen , Witold Pedrycz , Wei Wang , Fan Li","doi":"10.1016/j.compeleceng.2025.110252","DOIUrl":"10.1016/j.compeleceng.2025.110252","url":null,"abstract":"<div><div>Ensemble classification is an important branch and research focus in machine learning and pattern recognition. The current main paradigm of ensemble classification algorithms is based on same original samples, resulting in limited diversity between subsets. Therefore, it is particularly important to mine diverse and effective information from the original samples to build a multi-layer samples. However, samples are input into classifier for training one by one, or batch by batch in the existing ensemble algorithms. This ignores the potential value of correlation information among samples during base classifier training. To solve these problems, a new sample transformation model for ensemble classification - Hierarchical Manifold Sample Envelope Transformation Model (HMSET) is proposed. The model consists of three main parts. The first part is manifold sample enveloping model. It extracts local correlation among samples, thereby constructing manifold envelope samples. The second part is hierarchical sample envelope transformation model, which uses a variety of transformation operators and interlayer consistency to mine and constrain the correlation information among samples to enhance the diversity. The third part is two-dimensional fusion mechanism which fuse the final prediction results of the base classifiers. The 19 UCI datasets and several representative algorithms are used for validation. The results show that compared with the original samples, the proposed model improves the diversity of the sample subsets significantly. Compared with related ensemble classification algorithms, the proposed model has significantly better performance of ensemble classification.</div><div>Data and code are available in: <span><span>https://github.com/acceptthisjj/HMSET</span><svg><path></path></svg></span></div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"123 ","pages":"Article 110252"},"PeriodicalIF":4.0,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143738232","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Study on the Spacing between Movable Bus Stops and Signalized Intersections under Cooperative Vehicle Infrastructure Environment","authors":"Rui Li , Tianjing Qi , Xin Xue , Le Gu , Tao Chen","doi":"10.1016/j.compeleceng.2025.110278","DOIUrl":"10.1016/j.compeleceng.2025.110278","url":null,"abstract":"<div><div>Bus stops and signalized intersections are bottlenecks in urban road traffic, and their combined effect can exacerbate congestion in these areas. This study aims to address this bottleneck by proposing the use of a roadside movable bus stop within a cooperative vehicle infrastructure environment. Therefore, this paper studies the spacing between bus stops upstream and signalized intersections in a cooperative vehicle infrastructure environment. The study first analyzes the interaction between signal intersection and bus stop, then enhances the variable time-interval safety distance strategy through integration with the cooperative vehicle infrastructure environment. It then establishes vehicle operation rules for regular road sections, bus stop sections, and intersection areas and develops a simulation environment. Finally, the traffic flow characteristics under the combined configuration are analyzed, and reasonable recommendations are provided for setting the movable stop spacing upstream of intersections within the cooperative vehicle infrastructure environment. The results show that the connected vehicles can improve the traffic efficiency of the signalized intersection-stop section, and the intersection-stop spacing needs to be flexibly adjusted according to the traffic level, the penetration rate of connected vehicles, and bus dwell time. This paper provides a theoretical basis for urban bus stop planning in the cooperative vehicle infrastructure environment.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"123 ","pages":"Article 110278"},"PeriodicalIF":4.0,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143760344","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}