Journal of Information and Intelligence最新文献

{"title":"Advanced technologies for security, privacy, safety, and regulation of AI-generated content","authors":"Jianbing Ni, Cheng Huang, Wenchao Xu, Yue Zhang, Rongxing Lu","doi":"10.1016/j.jiixd.2026.06.005","DOIUrl":"10.1016/j.jiixd.2026.06.005","url":null,"abstract":"","PeriodicalId":100790,"journal":{"name":"Journal of Information and Intelligence","volume":"4 4","pages":"Pages 273-275"},"PeriodicalIF":0.0,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148666751","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Navigating the privacy-explainability-utility trilemma: A survey of differential privacy and explainable AI","authors":"Xiaowei Sun,&nbsp;Haichuan Xue,&nbsp;Dilli Prasad Sharma,&nbsp;Liang Xue,&nbsp;Xiaodong Lin,&nbsp;Pulei Xiong","doi":"10.1016/j.jiixd.2026.05.005","DOIUrl":"10.1016/j.jiixd.2026.05.005","url":null,"abstract":"<div><div>Artificial Intelligence (AI) systems used in sensitive domains must balance privacy protection and explainability. Differential Privacy (DP) provides formal privacy guarantees, while Explainable AI (XAI) aims to make model decisions understandable. However, privacy-preserving noise often degrades explanation quality, making their combination challenging. This survey provides a structured review of literature published between 2020 and 2025 and differs from prior threat-oriented or benchmark-driven surveys by adopting a mechanism-centered perspective that explicitly examines how DP techniques interact with different explanation paradigms. From this broader technical and system design perspective, we analyze DP and XAI interactions beyond purely threat-oriented considerations. We analyze these interactions through the lens of the privacy-explainability-utility trade-off, which shows that improving privacy can harm either model performance or explanation quality. To help organize and advance this growing field, we propose a five-category taxonomy that characterizes how DP and XAI connect in machine learning systems. We also introduce the Differential Privacy-Explainable AI (DP-XAI) interaction matrix, which maps combinations of privacy techniques and explanation methods that have been studied. Our analysis identifies several research gaps, especially in concept-based explanations and underused architectures, and highlights concrete directions for future research on building machine learning systems that are both privacy-protected and meaningfully explainable.</div></div>","PeriodicalId":100790,"journal":{"name":"Journal of Information and Intelligence","volume":"4 4","pages":"Pages 328-349"},"PeriodicalIF":0.0,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148666791","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"CRAFT: Complexity-routed attention with frequency-temporal fusion for multi-tab website fingerprinting","authors":"Yuhao Pan,&nbsp;Fushuo Huo,&nbsp;Haosong Peng,&nbsp;Kang Wei","doi":"10.1016/j.jiixd.2026.05.004","DOIUrl":"10.1016/j.jiixd.2026.05.004","url":null,"abstract":"<div><div>Tor is a popular low-latency anonymous communication system that protects user privacy via layered encryption. However, Website Fingerprinting (WF) attacks can still infer browsing behavior by analyzing statistical patterns in encrypted traffic. Existing multi-tab WF methods primarily rely on temporal features and employ fixed attention windows that treat all traffic segments uniformly, ignoring the fact that multi-tab traffic exhibits varying degrees of overlap across different time periods. To address these limitations, we propose CRAFT, a complexity-routed attention framework with frequency-temporal fusion for multi-tab website fingerprinting. First, we introduce a dual-branch feature extractor that combines a temporal convolutional backbone with a frequency-domain branch for capturing periodic traffic rhythms, fused via a learned gating mechanism. Second, we design a complexity-aware adaptive routed attention mechanism, where a lightweight router predicts per-token traffic mixing complexity and dynamically adjusts the attention scope through a Gaussian distance bias, focusing locally on simple single-site segments while expanding to broader context in complex overlapping regions. Extensive experiments across multiple multi-tab settings demonstrate that CRAFT achieves strong and consistent improvements over existing approaches.</div></div>","PeriodicalId":100790,"journal":{"name":"Journal of Information and Intelligence","volume":"4 4","pages":"Pages 298-310"},"PeriodicalIF":0.0,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148666753","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An information-driven and intelligent federated energy-timing watermark framework for man-in-the-middle attack detection in electric vehicle charging networks","authors":"Paritosh Biswas,&nbsp;Sushil Kumar Singh,&nbsp;Vinay Rishiwal","doi":"10.1016/j.jiixd.2026.04.005","DOIUrl":"10.1016/j.jiixd.2026.04.005","url":null,"abstract":"<div><div>The rapid adoption of Electric Vehicle Charging Networks (EVCNs) in smart city and IoT ecosystems has introduced new cyber-physical attack surfaces at the edge or physical layer, where charging sessions and control signals flow between Electric Vehicles (EVs) and charging stations. In these distributed environments, ensuring the authenticity and integrity of charging signals is critical for safety, billing, and grid stability. A pressing challenge for EVCNs is their susceptibility to Man-in-the-Middle (MitM) and relay attacks that intercept, delay, or modify charging signals to spoof vehicle responses or manipulate energy flows. Conventional digital defences-cryptographic protocols and signature-based detectors-often miss timing- and energy-based manipulations that emulate valid traffic patterns. This paper addresses that gap by proposing a hybrid cyber-physical detection framework that integrates Energy-Timing Watermarking (ETW) with a privacy-preserving federated learning engine (FedFox). ETW embeds subtle, verifiable perturbations into the charging energy/timing profile, creating a physical watermark that a legitimate EV response must reproduce a relay/MitM adversary cannot reliably mimic this physical fingerprint. FedFox enables collaborative, decentralized model training across charging stations without sharing raw telemetry, improving detection robustness while preserving user privacy. We evaluate the approach on a proxy IoT intrusion dataset adapted for MitM scenarios and simulate federated rounds with ETW client validation. The proposed ETW-FedFox framework is designed specifically to target relay and MitM attacks, and, when evaluated on the IoTID20 proxy dataset with synthetic ETW-like features, achieves approximately 99.7% accuracy and 99.85% F1-score with low false-positive rates. These results demonstrate that, under the assumed synthetic feature model, combining lightweight physical-layer-inspired watermarking with federated learning can effectively separate benign and malicious traffic. This paper introduces the ETW-FedFox architecture and proves its ability to detect anomalies based on the IoTID20 proxy dataset, which simulates EVCS-like IoT traffic. While the ETW design is physical-layer-oriented, the current evaluation does not yet include a physically connected EV battery/BMS; real EVCS experiments are left as important future work.</div></div>","PeriodicalId":100790,"journal":{"name":"Journal of Information and Intelligence","volume":"4 4","pages":"Pages 311-327"},"PeriodicalIF":0.0,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148666749","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"From AI-generated content to agentic action: Security and safety threats in generative AI","authors":"Zelin Zhang,&nbsp;Qi Li,&nbsp;Jie Cao,&nbsp;Lingshuang Liu,&nbsp;Jianbing Ni","doi":"10.1016/j.jiixd.2026.05.002","DOIUrl":"10.1016/j.jiixd.2026.05.002","url":null,"abstract":"<div><div>Generative AI systems are increasingly used not only to produce content but also to retrieve data, invoke tools, and execute actions. This work examines the security and safety implications of that shift across content-level, model-level, and agentic threats. We analyze how attacker access requirements, system autonomy, and the scope of potential harm change as models move from generating artifacts to executing operations through tool chains and external APIs. We then assess technical countermeasures including detection, watermarking, alignment, and emerging agentic safeguards, and show that several depend on forms of institutional coordination that current governance arrangements do not yet provide. Across the cases examined, capability deployment and attack-surface expansion repeatedly outpace defensive responses as systems move from generating content to executing real-world actions.</div></div>","PeriodicalId":100790,"journal":{"name":"Journal of Information and Intelligence","volume":"4 4","pages":"Pages 276-297"},"PeriodicalIF":0.0,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148666752","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Privacy-preserving electricity load forecasting using hierarchical federated learning","authors":"Zeyu He,&nbsp;Haonan Zhang,&nbsp;Yining Liu","doi":"10.1016/j.jiixd.2026.04.009","DOIUrl":"10.1016/j.jiixd.2026.04.009","url":null,"abstract":"<div><div>Smart meters, utilizing their large-scale distributed deployment architecture, facilitate real-time acquisition of fine-grained residential load data, which significantly enhances the accuracy of Short-term Residential electrical Load Forecasting (SRLF) models. However, SRLF typically requires fine-grained user data as input, which raises privacy concerns. Moreover, smart meters, as computing resource-constrained devices, are unable to perform complex training processes locally. Furthermore, the large-scale deployment of smart meters imposes significant communication overhead and latency on networks. To address these challenges, we propose an end-edge-cloud hierarchical Federated Learning (FL) framework that leverages fine-grained residential load data from numerous smart meters in a privacy-preserving and resource-efficient manner. Specifically, the primary training and prediction tasks are shifted to edge servers proximate to the smart meters. An inner-product functional encryption scheme is employed, enabling local training and prediction to be performed by the edge server for multiple smart meters without their raw data being accessed. Furthermore, a hierarchical aggregation strategy ensures the equitable and comprehensive utilization of distributed data. Extensive experiments demonstrate our scheme achieving better performance in the practical environment.</div></div>","PeriodicalId":100790,"journal":{"name":"Journal of Information and Intelligence","volume":"4 4","pages":"Pages 350-363"},"PeriodicalIF":0.0,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148666792","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A survey on bias and fairness in machine unlearning","authors":"Mohammad Partohaghighi ,&nbsp;Roummel Marcia ,&nbsp;Bruce J. West ,&nbsp;YangQuan Chen","doi":"10.1016/j.jiixd.2026.03.003","DOIUrl":"10.1016/j.jiixd.2026.03.003","url":null,"abstract":"<div><div>Machine Unlearning (MU) aims to remove the influence of specified training data from a deployed model, motivated by privacy regulation (e.g., General Data Protection Regulation (GDPR) right to erasure), safety (e.g., removing harmful or poisoned data), and governance (e.g., policy-driven model updates). While MU has advanced from exact/sharded retraining to certified and approximate unlearning with verification mechanisms, its interaction with bias and fairness remains substantially less systematized than classical algorithmic fairness. This survey synthesizes the emerging literature on bias and fairness in MU. We (1) use a data–model–user–deletion–verification loop as an organizing lens to highlight new bias channels created by deletion policies and auditing; (2) introduce a survey-level organizing vocabulary for unlearning-specific fairness notions, including deletion<em>-</em>selection bias, retention fairness, and an expanded fairness<em>-</em>of<em>-</em>forgetting framework; (3) provide a dedicated threat model and stakeholder incentives analysis; (4) catalog benchmarks, datasets, and deletion protocols needed for cblackible fairness conclusions; (5) broaden fairness metrics beyond binary classification to regression, ranking/recommendation, and generative/LLM unlearning; and (6) provide practitioner-facing deployment and reporting guidance, including group-conditioned fidelity and verification error rates. This article is primarily intended as a survey of the emerging literature on fairness in machine unlearning. Where prior work remains conceptually fragmented, we additionally introduce a survey-level synthesis and organizing vocabulary to consolidate recurring phenomena, rather than to claim universally standardized terminology.</div></div>","PeriodicalId":100790,"journal":{"name":"Journal of Information and Intelligence","volume":"4 3","pages":"Pages 183-201"},"PeriodicalIF":0.0,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148183587","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Adaptive multi-phased wear leveling method for SSD lifetime enhancement","authors":"Tingyu Wang ,&nbsp;Mengfei Yang ,&nbsp;Jingjing Jiang ,&nbsp;Hongbiao Liu ,&nbsp;Shaofeng Li ,&nbsp;Zhi Ma","doi":"10.1016/j.jiixd.2025.10.001","DOIUrl":"10.1016/j.jiixd.2025.10.001","url":null,"abstract":"<div><div>In NAND flash-based Solid-State Drives (SSDs), the limited endurance of flash memory makes Wear Leveling (WL) a critical technique for extending the lifetime and improving reliability. However, existing WL methods rely on fixed thresholds or a single dynamic threshold applied uniformly throughout SSD lifetime, which fails to adapt to dynamic wear conditions, such as uneven block wear and workload-induced variations in I/O intensity. This limitation can result in excessive data migration and performance degradation. This paper proposes an adaptive multi-phased wear leveling method (called AMPWL) to extend SSD lifetime and improve system performance. Firstly, AMPWL divides the SSD lifetime into three distinct phases to better capture its wear characteristics. Secondly, phase-specific dynamic thresholds are designed to adaptively trigger wear leveling operations. Thirdly, a scoring mechanism is introduced to enhance wear leveling efficiency. Experimental results demonstrate that AMPWL significantly outperforms existing wear leveling methods in extending SSD lifetime and improving I/O performance.</div></div>","PeriodicalId":100790,"journal":{"name":"Journal of Information and Intelligence","volume":"4 3","pages":"Pages 202-220"},"PeriodicalIF":0.0,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148183588","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A multi-objective scheme for collision avoidance, swarm cohesion, and target tracking for smart UAVs","authors":"Jawad Mahmood,&nbsp;Muhammad Adil Raja,&nbsp;John Loane,&nbsp;Fergal McCaffery","doi":"10.1016/j.jiixd.2025.10.004","DOIUrl":"10.1016/j.jiixd.2025.10.004","url":null,"abstract":"<div><div>This research introduces a Reinforcement Learning (RL) based testbed designed for real-time coordination among multiple Unmanned Aerial Vehicles (UAVs) using actor-critic models to control a randomly manoeuvring target UAV, and a swarm of tracking UAVs. The system simulates realistic multi-agent interactions within a scalable and distributed architecture built on FlightGear and JSBSim. A key feature is the ability of multiple agents to learn independently yet collaboratively, enabling synchronized behaviors and adaptive tracking in dynamic environments. Collision avoidance, swarm cohesion, and target tracking efficiency are ensured through a 3D ellipsoidal spatial model and a Gaussian reward function, which guide agents to maintain optimal relative positions. The integration of an Intrinsic Curiosity Module (ICM) enhances exploration in sparse-reward settings. Experimental results show that the multi-agent swarm UAVs effectively tracks complex and evasive target paths, demonstrating robustness, scalability, and real-time coordination capabilities essential for advanced UAV operations.</div></div>","PeriodicalId":100790,"journal":{"name":"Journal of Information and Intelligence","volume":"4 3","pages":"Pages 221-235"},"PeriodicalIF":0.0,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148183049","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"HTA-BotDef: A novel hierarchical tree aggregation for scalable and privacy preserving botnet attack detection with federated learning","authors":"Md. Alamgir Hossain ,&nbsp;Md. Samiul Islam","doi":"10.1016/j.jiixd.2025.12.004","DOIUrl":"10.1016/j.jiixd.2025.12.004","url":null,"abstract":"<div><div>The accelerating sophistication and diversity of botnet attacks continue to undermine network resilience, necessitating defense frameworks that are both decentralized and privacy-preserving. This study presents HTA-BotDef, a novel Hierarchical Tree Aggregation (HTA) framework designed for scalable and secure botnet detection through Federated Learning (FL). Unlike conventional centralized paradigms, HTA-BotDef enables distributed clients to collaboratively train a global detection model without exchanging raw data, thereby preserving privacy while maintaining high detection efficacy. To emulate real-world, non-IID network conditions, the framework integrates heterogeneous datasets including N-BaIoT, ISCX, MedBIoT, and NCC2, encompassing diverse botnet families such as Mirai, Bashlite, IRC Bot, Torii, Neris, and Rbot. The proposed architecture combines advanced feature selection strategies, correlation analysis, mutual information, and Principal Component Analysis (PCA) with Random Forest-based local classifiers, aggregated via the hierarchical tree mechanism to form an adaptive and resilient global model. Dynamic hyperparameter optimization further enhances cross-client generalization and convergence stability. Extensive experiments demonstrate consistent and near-perfect performance, with accuracy, precision, and recall values exceeding 99.99% even under severe class imbalance. The findings establish HTA-BotDef as a highly scalable and privacy-preserving solution for real-time botnet detection, offering a significant advancement toward trustworthy and decentralized network defense systems.</div></div>","PeriodicalId":100790,"journal":{"name":"Journal of Information and Intelligence","volume":"4 3","pages":"Pages 251-271"},"PeriodicalIF":0.0,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148183590","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}