Jianbo Ding
Research Subjects
Jianbo Ding has a total of 10 co-authors. Here are the top co-authors by number of shared publications:
- Fengyuan Zhang
Tandon School of Engineering, New York University, New York, NY 10010, USA
2 shared publications - Wenbin Shang
University of Glasgow, Glasgow, U.K.
2 shared publications - Wenhe Liu
School of Computer Science, Carnegie Mellon University, Pittsburgh, PA 15213, USA
2 shared publications - Dai Teng
University of Illinois at Urbana–Champaign
1 shared publication - Jie Si Yang
The University of Utah, Salt Lake City, UT, USA
1 shared publication - Jingyun Yang
David A. Tepper School of Business, Carnegie Mellon University, Pittsburgh, PA 15213, USA
1 shared publication
Latest Publications
- Open Access
- 19 August 2026
- Published by Elsevier in SSRN Electronic Journal
Adaptive Zero Trust Access Control Using Explainable Risk-Aware Machine Learning for Multi-Tenant Cloud Environments
Zero Trust Architecture requires continuous, context-dependent authorization, yet a learned risk score is not itself an access-control policy. Fixed score thresholds can create large cross-tenant false-denial disparities and can fail under temporal score shift, while black-box hard denials are difficult to justify operationally. This paper presents HTEC-ZT (Hierarchical Tenant-Calibrated Explainable Zero Trust), a concrete three-action access-control algorithm that combines: (i) dual LightGBM risk estimators with disagreement-aware risk inflation; (ii) empirical-Bayes-style hierarchical benign-risk quantiles for tenant-specific step-up and hard-deny budgets; (iii) a Counterfactual Explanation Agreement (CEA) gate that downgrades a hard denial to step-up authentication when two independently trained models disagree on the locally risk-reducing features; and (iv) delayed verified-benign threshold adaptation that excludes the current decision batch to prevent label leakage. Experiments use two downloadable public datasets: an AWS-hosted enterprise scenario from CSE-CIC-IDS2018 and the official UNSW-NB15 split. Because neither flow table contains a native cloud tenant identifier, tenant partitions are deliberately emulated using label-independent networking attributes and are reported as a controlled experimental construct. On UNSW-NB15, the CEA gate reduces benign hard-deny rate from 13.01% to 6.85% while preserving 98.68% attack intervention coverage. On the temporally shifted CSE-CIC-IDS2018 stream, it reduces benign hard-deny from 41.21% to 17.10% at unchanged 74.26% intervention coverage; delayed adaptation further reduces benign hard-deny to 6.05% and tenant-rate standard deviation from 0.122 to 0.025, with a 1.41 percentage-point intervention trade-off. Batch measurements on a five-vCPU container show a CEA amortized cost of 0.066-0.087 ms per hard-deny candidate. The results establish an auditable, reproducible path from anomaly scores to adaptive multi-tenant access decisions rather than a broad Zero Trust framework.
- 19 August 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessAdaptive Zero Trust Access Control Using Explainable Risk-Aware Machine Learning for Multi-Tenant Cloud Environments
Zero Trust Architecture requires continuous, context-dependent authorization, yet a learned risk score is not itself an access-control policy. Fixed score thresholds can create large cross-tenant false-denial disparities and can fail under temporal score shift, while black-box hard denials are difficult to justify operationally. This paper presents HTEC-ZT (Hierarchical Tenant-Calibrated Explainable Zero Trust), a concrete three-action access-control algorithm that combines: (i) dual LightGBM risk estimators with disagreement-aware risk inflation; (ii) empirical-Bayes-style hierarchical benign-risk quantiles for tenant-specific step-up and hard-deny budgets; (iii) a Counterfactual Explanation Agreement (CEA) gate that downgrades a hard denial to step-up authentication when two independently trained models disagree on the locally risk-reducing features; and (iv) delayed verified-benign threshold adaptation that excludes the current decision batch to prevent label leakage. Experiments use two downloadable public datasets: an AWS-hosted enterprise scenario from CSE-CIC-IDS2018 and the official UNSW-NB15 split. Because neither flow table contains a native cloud tenant identifier, tenant partitions are deliberately emulated using label-independent networking attributes and are reported as a controlled experimental construct. On UNSW-NB15, the CEA gate reduces benign hard-deny rate from 13.01% to 6.85% while preserving 98.68% attack intervention coverage. On the temporally shifted CSE-CIC-IDS2018 stream, it reduces benign hard-deny from 41.21% to 17.10% at unchanged 74.26% intervention coverage; delayed adaptation further reduces benign hard-deny to 6.05% and tenant-rate standard deviation from 0.122 to 0.025, with a 1.41 percentage-point intervention trade-off. Batch measurements on a five-vCPU container show a CEA amortized cost of 0.066-0.087 ms per hard-deny candidate. The results establish an auditable, reproducible path from anomaly scores to adaptive multi-tenant access decisions rather than a broad Zero Trust framework.
- Top Cited PapersOpen Access
Adaptive Sensor Fusion for Robust Perception in Dense Fog: A Gated Vision and LiDAR Integration Framework
Autonomous driving systems face critical perception failures in dense fog, where conventional RGB cameras suffer from severe degradation due to atmospheric scattering and reduced visibility. This paper presents an adaptive multi-modal fusion framework that synergistically integrates gated imaging with 3D LiDAR point clouds to achieve robust obstacle detection under visibility conditions as low as 50 m. Unlike standard cameras that passively capture scattered ambient light, gated cameras employ time-synchronized active illumination to physically filter backscattered photons, preserving structural features even in low-visibility scenarios. We propose a novel Adaptive Feature-Weighting Network (AFW-Net) that dynamically adjusts sensor modality contributions based on real-time environmental degradation assessment. The framework incorporates three key innovations: (1) a cross-modal feature extraction module that exploits the complementary physical properties of gated imaging and LiDAR, (2) an attention-based adaptive fusion mechanism that quantifies per-modality reliability through uncertainty estimation, and (3) a degradation-aware training strategy using weather-specific augmentation. Extensive experiments on the Princeton Automated Driving Dataset demonstrate that our approach maintains detection average precision (AP) above 82% under dense fog conditions (50 m visibility), representing a 23.7% improvement over state-of-the-art RGB-LiDAR fusion methods that exhibit substantial performance degradation to 58.4% AP. Ablation studies validate the necessity of each component, and cross-dataset evaluation confirms the generalization capability of the proposed framework. The adaptive weighting mechanism proves particularly effective, dynamically rebalancing modality contributions across the gated imaging and LiDAR branches while maintaining LiDAR geometric constraints. This work establishes a robust perception paradigm for safety-critical autonomous systems operating in low-visibility environmental conditions.
Top Cited PapersOpen AccessAdaptive Sensor Fusion for Robust Perception in Dense Fog: A Gated Vision and LiDAR Integration Framework
Autonomous driving systems face critical perception failures in dense fog, where conventional RGB cameras suffer from severe degradation due to atmospheric scattering and reduced visibility. This paper presents an adaptive multi-modal fusion framework that synergistically integrates gated imaging with 3D LiDAR point clouds to achieve robust obstacle detection under visibility conditions as low as 50 m. Unlike standard cameras that passively capture scattered ambient light, gated cameras employ time-synchronized active illumination to physically filter backscattered photons, preserving structural features even in low-visibility scenarios. We propose a novel Adaptive Feature-Weighting Network (AFW-Net) that dynamically adjusts sensor modality contributions based on real-time environmental degradation assessment. The framework incorporates three key innovations: (1) a cross-modal feature extraction module that exploits the complementary physical properties of gated imaging and LiDAR, (2) an attention-based adaptive fusion mechanism that quantifies per-modality reliability through uncertainty estimation, and (3) a degradation-aware training strategy using weather-specific augmentation. Extensive experiments on the Princeton Automated Driving Dataset demonstrate that our approach maintains detection average precision (AP) above 82% under dense fog conditions (50 m visibility), representing a 23.7% improvement over state-of-the-art RGB-LiDAR fusion methods that exhibit substantial performance degradation to 58.4% AP. Ablation studies validate the necessity of each component, and cross-dataset evaluation confirms the generalization capability of the proposed framework. The adaptive weighting mechanism proves particularly effective, dynamically rebalancing modality contributions across the gated imaging and LiDAR branches while maintaining LiDAR geometric constraints. This work establishes a robust perception paradigm for safety-critical autonomous systems operating in low-visibility environmental conditions.
- Top Cited PapersOpen Access
- 12 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 72891-72904
FairPromote: Explainable and Fairness-Aware Talent Promotion Prediction via Adversarial Debiasing and SHAP-Based Interpretation
Talent promotion decisions are critical for organizational success, yet traditional prediction models often suffer from opacity and inherited biases embedded in historical data. These black-box systems not only lack transparency but also perpetuate discriminatory patterns related to sensitive attributes such as gender and geographic location, undermining organizational fairness and employee trust. Moreover, existing evaluation metrics struggle to quantify the unstructured contributions of senior management, leading to poor generalization in complex personnel decisions. To address these challenges, we propose FairPromote, a novel framework that integrates SHAP (SHapley Additive exPlanations) values with adversarial debiasing techniques within an ensemble learning architecture. Our approach simultaneously achieves high prediction accuracy while automatically identifying and mitigating the negative impact of sensitive attributes on promotion decisions. The framework employs a dual-objective optimization strategy: a primary predictor maximizes promotion prediction performance, while an adversarial discriminator minimizes the ability to infer sensitive attributes from learned representations. Through SHAP-based local explanation plots, FairPromote provides instance-level attribution analysis for each personnel decision, addressing the critical trust deficit in algorithmic decision-making systems. We evaluate our framework on the IBM HR Analytics dataset, demonstrating superior performance across multiple dimensions. Experimental results show that FairPromote achieves 91.3% prediction accuracy while reducing demographic parity difference by 68% and equalized odds disparity by 72% compared to baseline models. The SHAP-based interpretations reveal that job involvement, years at company, and performance ratings are the primary drivers of promotion decisions, while successfully neutralizing the influence of gender and marital status.
- 12 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 72891-72904
Top Cited PapersOpen AccessFairPromote: Explainable and Fairness-Aware Talent Promotion Prediction via Adversarial Debiasing and SHAP-Based Interpretation
Talent promotion decisions are critical for organizational success, yet traditional prediction models often suffer from opacity and inherited biases embedded in historical data. These black-box systems not only lack transparency but also perpetuate discriminatory patterns related to sensitive attributes such as gender and geographic location, undermining organizational fairness and employee trust. Moreover, existing evaluation metrics struggle to quantify the unstructured contributions of senior management, leading to poor generalization in complex personnel decisions. To address these challenges, we propose FairPromote, a novel framework that integrates SHAP (SHapley Additive exPlanations) values with adversarial debiasing techniques within an ensemble learning architecture. Our approach simultaneously achieves high prediction accuracy while automatically identifying and mitigating the negative impact of sensitive attributes on promotion decisions. The framework employs a dual-objective optimization strategy: a primary predictor maximizes promotion prediction performance, while an adversarial discriminator minimizes the ability to infer sensitive attributes from learned representations. Through SHAP-based local explanation plots, FairPromote provides instance-level attribution analysis for each personnel decision, addressing the critical trust deficit in algorithmic decision-making systems. We evaluate our framework on the IBM HR Analytics dataset, demonstrating superior performance across multiple dimensions. Experimental results show that FairPromote achieves 91.3% prediction accuracy while reducing demographic parity difference by 68% and equalized odds disparity by 72% compared to baseline models. The SHAP-based interpretations reveal that job involvement, years at company, and performance ratings are the primary drivers of promotion decisions, while successfully neutralizing the influence of gender and marital status.
- Top Cited PapersOpen Access
- 18 April 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (8) , 3944
Game-Theoretic Cost-Sensitive Adversarial Training for Robust Cloud Intrusion Detection Against GAN-Based Evasion Attacks
Cloud-based intrusion detection systems (IDSs) increasingly rely on deep learning classifiers to identify malicious traffic; however, this reliance exposes them to adversarial evasion attacks in which adversaries craft near-imperceptible perturbations to bypass detection. Existing defenses based on conventional adversarial training often recover robustness against known perturbation patterns at the cost of degraded detection accuracy on canonical attack categories—a robustness–accuracy trade-off that remains an open challenge in the field. In this paper, we propose GT-CSAT (Game-Theoretic Cost-Sensitive Adversarial Training), a novel defense framework tailored for cloud security environments. GT-CSAT couples an improved Wasserstein GAN with Gradient Penalty (WGAN-GP) threat generator—conditioned on attack semantics to simulate functionally consistent and highly covert traffic variants—with a minimax adversarial training loop governed by a game-theoretic cost-sensitive loss function. The proposed loss function assigns asymmetric misclassification penalties derived from a two-player zero-sum payoff matrix, enabling the detector to maintain vigilance over both novel adversarial variants and well-characterized conventional threats simultaneously. Specifically, misclassifying an adversarially perturbed attack as benign incurs a strictly higher penalty than the symmetric cross-entropy baseline, while the cost weights are dynamically adapted via a Nash equilibrium-inspired update rule during training. We conduct comprehensive experiments on the Cloud Vulnerabilities Dataset (CVD), CICIDS-2017, and UNSW-NB15, which encompass diverse cloud-specific attack scenarios including denial-of-service, port scanning, brute-force, and SQL injection traffic. Under six representative evasion strategies—FGSM, PGD, C&W, BIM, DeepFool, and IDSGAN-style black-box perturbations—GT-CSAT achieves an average robust accuracy of 94.3%, surpassing standard adversarial training by 6.8 percentage points and the undefended baseline by 21.4 percentage points, while preserving clean-traffic detection at 97.1%. These results confirm that the game-theoretic cost structure effectively decouples robustness from accuracy, yielding a Pareto-superior detection profile relative to competing baselines across all evaluated threat models. The source code and experimental configurations have been publicly released to facilitate reproducibility.
- 18 April 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (8) , 3944
Top Cited PapersOpen AccessGame-Theoretic Cost-Sensitive Adversarial Training for Robust Cloud Intrusion Detection Against GAN-Based Evasion Attacks
Cloud-based intrusion detection systems (IDSs) increasingly rely on deep learning classifiers to identify malicious traffic; however, this reliance exposes them to adversarial evasion attacks in which adversaries craft near-imperceptible perturbations to bypass detection. Existing defenses based on conventional adversarial training often recover robustness against known perturbation patterns at the cost of degraded detection accuracy on canonical attack categories—a robustness–accuracy trade-off that remains an open challenge in the field. In this paper, we propose GT-CSAT (Game-Theoretic Cost-Sensitive Adversarial Training), a novel defense framework tailored for cloud security environments. GT-CSAT couples an improved Wasserstein GAN with Gradient Penalty (WGAN-GP) threat generator—conditioned on attack semantics to simulate functionally consistent and highly covert traffic variants—with a minimax adversarial training loop governed by a game-theoretic cost-sensitive loss function. The proposed loss function assigns asymmetric misclassification penalties derived from a two-player zero-sum payoff matrix, enabling the detector to maintain vigilance over both novel adversarial variants and well-characterized conventional threats simultaneously. Specifically, misclassifying an adversarially perturbed attack as benign incurs a strictly higher penalty than the symmetric cross-entropy baseline, while the cost weights are dynamically adapted via a Nash equilibrium-inspired update rule during training. We conduct comprehensive experiments on the Cloud Vulnerabilities Dataset (CVD), CICIDS-2017, and UNSW-NB15, which encompass diverse cloud-specific attack scenarios including denial-of-service, port scanning, brute-force, and SQL injection traffic. Under six representative evasion strategies—FGSM, PGD, C&W, BIM, DeepFool, and IDSGAN-style black-box perturbations—GT-CSAT achieves an average robust accuracy of 94.3%, surpassing standard adversarial training by 6.8 percentage points and the undefended baseline by 21.4 percentage points, while preserving clean-traffic detection at 97.1%. These results confirm that the game-theoretic cost structure effectively decouples robustness from accuracy, yielding a Pareto-superior detection profile relative to competing baselines across all evaluated threat models. The source code and experimental configurations have been publicly released to facilitate reproducibility.
- Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 111168-111187
Neuro-Elastic: A Unified Framework for Hardware-Aware Adaptive Quantization and Dynamic Sparsity in Real-Time Edge Intent Prediction
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 111168-111187
Open AccessNeuro-Elastic: A Unified Framework for Hardware-Aware Adaptive Quantization and Dynamic Sparsity in Real-Time Edge Intent Prediction