Journal of Reliable Intelligent Environments最新文献

Privacy-aware adversarial defense with explainable AI for adversarial robustness in AI model. 具有可解释AI的隐私感知对抗防御，用于AI模型中的对抗鲁棒性。

Journal of Reliable Intelligent Environments Pub Date : 2026-01-01 Epub Date: 2026-09-15 DOI: 10.1007/s40860-026-00273-7

Bilal Sardar, Shareeful Islam, Stefano Silvestri, Spyridon Papastergiou

{"title":"Privacy-aware adversarial defense with explainable AI for adversarial robustness in AI model.","authors":"Bilal Sardar, Shareeful Islam, Stefano Silvestri, Spyridon Papastergiou","doi":"10.1007/s40860-026-00273-7","DOIUrl":"10.1007/s40860-026-00273-7","url":null,"abstract":"AI models in critical sectors such as healthcare and finance must provide both data privacy and adversarial robustness. Differential Privacy (DP) protects training data by injecting noise, but this noise smooths decision boundaries and leaves models open to adversarial evasion, a tension known as the Privacy-Robustness Trade-off. Although this trade-off is well documented, its internal mechanism remains underexplored: prior work does not reveal how the noise reshapes a model's reasoning or which features become vulnerable. To close this gap, we propose a Privacy-Aware Adversarial Defense grounded in Explainable AI. Specifically, we introduce the Attention Concentration Score (ACS), which measures how DP training shifts Transformer attention away from task-critical features toward non-functional ones. This attention drift correlates with adversarial vulnerability, providing a mechanistic explanation of the trade-off. Building on this insight, we develop a Manifold-Aligned Semantic Attack that targets the most drifted features, and a TrustScore defense that fuses embedding-level anomaly detection with attention-level consistency checks. We validate across two datasets (Adult Census, MIMIC-IV), two architectures (DeBERTa-V3-Large, LLaMA-3.1-8B), and seven experiments benchmarking five attacks against six defenses. Within the recommended range ( <math><mrow><mi>ϵ</mi> <mo>∈</mo> <mo>[</mo> <mn>5</mn> <mo>,</mo> <mn>10</mn> <mo>]</mo></mrow> </math> ), models retain 84.2% accuracy (96.2% of baseline), while TrustScore reaches an Area Under the ROC Curve (AUC) of 0.87--0.94, outperforming Isolation Forest (0.65) and supervised detection (0.58). Moreover, these conclusions hold under feature-categorization variants (attribution- and PCA-based); privacy noise disproportionately destabilizes the minority class; the consistency signal adds sub-millisecond overhead; and a surrogate-attention variant preserves detection under black-box deployment, establishing the approach's dependability for reliable intelligent environments.","PeriodicalId":37011,"journal":{"name":"Journal of Reliable Intelligent Environments","volume":"12 4","pages":"13"},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13577975/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148983603","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}

引用次数: 0

Intelligent dynamic cybersecurity risk management framework with explainability and interpretability of AI models for enhancing security and resilience of digital infrastructure. 智能动态网络安全风险管理框架，具有可解释性和可解释性的人工智能模型，用于增强数字基础设施的安全性和弹性。

Journal of Reliable Intelligent Environments Pub Date : 2025-01-01 Epub Date: 2025-07-09 DOI: 10.1007/s40860-025-00253-3

Shareeful Islam, Nihala Basheer, Spyridon Papastergiou, Mario Ciampi, Stefano Silvestri

{"title":"Intelligent dynamic cybersecurity risk management framework with explainability and interpretability of AI models for enhancing security and resilience of digital infrastructure.","authors":"Shareeful Islam, Nihala Basheer, Spyridon Papastergiou, Mario Ciampi, Stefano Silvestri","doi":"10.1007/s40860-025-00253-3","DOIUrl":"10.1007/s40860-025-00253-3","url":null,"abstract":"The sophistication of cyberattacks has significantly increased, making it almost certain that organizations can be victims of cyberattacks at any time. Managing cybersecurity risk is critical for any organization so that informed decisions can be made to tackle risks before they materialize. Cybersecurity risk management is context-specific and heavily relies on the specific organization's context. However, performing effective risk management is always challenging due to the constant changes in organizational infrastructure and security posture, including the adoption of new applications and the reconfiguration or updating of existing assets and their dependencies, as well as the potential exploitation of vulnerabilities. Despite the wider adoption of AI enabled cybersecurity risk management, there is a lack of focus on the integration of these systems along with the dynamic elements of the risk management. In this context, this research proposes a novel dynamic cyber security risk management (d-CSRM) framework to tackle this challenge by integrating dynamic parameters such as vulnerability exploitation and assets dependencies for assessing and managing the risk. The framework consists of a systemic process and makes use of a hybrid AI-enabled model that combines both linear regression and deep learning, to prioritize the vulnerabilities. Additionally, d-CSRM integrates the explainability and interpretability characteristics of the AI model for explaining model decision making and the inner working parameters. This allows the extraction of the key features that are linked with the risk and informed decision making to tackle the risks. An experiment was performed to prioritize the vulnerabilities from the widely used CVEjoin dataset using the proposed hybrid model to quantify the dynamic risk with explainability. The results show that the hybrid model effectively identifies and prioritizes the most critical vulnerabilities using the selected key features such as exploit type, exploit platform and impact that can further enhance the dynamic risk assessment.","PeriodicalId":37011,"journal":{"name":"Journal of Reliable Intelligent Environments","volume":"11 3","pages":"12"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12241235/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144627360","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}

引用次数: 0

Development methodologies for IoT-based systems: challenges and research directions 基于物联网系统的开发方法：挑战与研究方向

Journal of Reliable Intelligent Environments Pub Date : 2024-08-09 DOI: 10.1007/s40860-024-00229-9

Miguel J. Hornos, Mario Quinde

引用次数: 0

A survey on graph neural network-based next POI recommendation for smart cities 基于图神经网络的智慧城市下一个 POI 推荐研究

Journal of Reliable Intelligent Environments Pub Date : 2024-07-26 DOI: 10.1007/s40860-024-00233-z

Jian Yu, Lucas Guo, Jiayu Zhang, Guiling Wang

引用次数: 0

Surveying neuro-symbolic approaches for reliable artificial intelligence of things 为可靠的物联网人工智能探索神经符号方法

Journal of Reliable Intelligent Environments Pub Date : 2024-07-26 DOI: 10.1007/s40860-024-00231-1

Zhen Lu, Imran Afridi, Hong Jin Kang, Ivan Ruchkin, Xi Zheng

引用次数: 0

Secure electronic monitoring of sex offenders 对性犯罪者进行安全的电子监控

Journal of Reliable Intelligent Environments Pub Date : 2024-07-11 DOI: 10.1007/s40860-024-00227-x

Francesco Buccafurri, V. de Angelis, Maria Francesca Idone, Cecilia Labrini

引用次数: 0

An efficient approach of epilepsy seizure alert system using IoT and machine learning 利用物联网和机器学习的高效癫痫发作警报系统

Journal of Reliable Intelligent Environments Pub Date : 2024-07-11 DOI: 10.1007/s40860-024-00228-w

Jagadeesh Basavaiah, Audre Arlene Anthony, S. Mahadevaswamy, H. N. Naveen Kumar

引用次数: 0

Dependability analysis and disaster recovery measures in smart hospital systems 智能医院系统的可靠性分析和灾难恢复措施

Journal of Reliable Intelligent Environments Pub Date : 2024-05-13 DOI: 10.1007/s40860-024-00222-2

Luiz Nelson Lima, Arthur Sabino, Vandirleya Barbosa, Leonel Feitosa, Carlos Brito, Jean Araujo, Francisco Airton Silva

引用次数: 0

Automated phase-type distribution fitting via expectation maximization 通过期望最大化实现自动相型分布拟合

Journal of Reliable Intelligent Environments Pub Date : 2024-04-24 DOI: 10.1007/s40860-024-00220-4

M. Mialaret, Paulo Pereira, Antônio Sá Barreto, Thiago Pinheiro, Paulo Maciel

引用次数: 0

Early fire danger monitoring system in smart cities using optimization-based deep learning techniques with artificial intelligence 使用基于优化的人工智能深度学习技术的智慧城市早期火险监测系统

Journal of Reliable Intelligent Environments Pub Date : 2024-03-29 DOI: 10.1007/s40860-024-00218-y

P. D. K. Reddy, Martin Margala, S. S. Shankar, Prasun Chakrabarti

引用次数: 0