Publications
Peer-reviewed research in Explainable AI, Video Understanding, and Cloud AI Systems
Top Tier Publications
4 papersA* ranked conferences and Q1 journals with high impact factors
STAA: Spatio-Temporal Attention Attribution for Real-Time Interpreting Transformer-Based AI Video Models
Zerui Wang, Yan Liu
IEEE Access
A novel XAI method for real-time interpretation of video Transformer models. STAA extracts spatio-temporal explanations in a single forward pass with less than 3% computational overhead, achieving superior faithfulness (0.87) and monotonicity (0.91) scores.
XAIport: A Service Framework for the Early Adoption of XAI in AI Model Development
Zerui Wang, Yan Liu, A. A. Thiruselvi, Wahab Hamou-Lhadj
46th IEEE/ACM International Conference on Software Engineering (ICSE)
A microservice framework for integrating Explainable AI into ML development pipelines, enabling early XAI adoption in MLOps workflows. Compatible with Azure, GCP, and AWS cloud services.
An Open API Architecture to Discover the Trustworthy Explanation of Cloud AI Services
Zerui Wang, Yan Liu, Jiahao Huang
IEEE Transactions on Cloud Computing
A cloud-agnostic architecture enabling explainability for proprietary cloud AI services (Azure, Google Cloud, AWS) without accessing internal model parameters.
Joint Spatio-temporal Adversarial Attacks on Video Transformer Models Through XAI-guided Gradient Perturbation
Zerui Wang, Yan Liu
ACM Transactions on Multimedia Computing, Communications, and Applications (TOMM)
First method to simultaneously perturb spatial and temporal features in video transformers. Achieves highest Attack Success Rate on Kinetics-400, revealing systematic security vulnerabilities.
High Quality Publications
7 papersB-ranked conferences and reputable venues
Cloud-Based XAI Services for Assessing Open Repository Models Under Adversarial Attacks
Zerui Wang, Yan Liu
IEEE International Conference on Software Services Engineering (SSE)
Comprehensive assessment protocol for evaluating open-source models including adversarial robustness, explanation deviation, and explanation resilience.
XAIpipeline: Automated Orchestration of Explainable AI Service for Cloud AI and Open-Source Models
Zerui Wang, Yan Liu
IEEE International Conference on Software Services Engineering (SSE)
Automated toolchain for orchestrating XAI services across cloud and open-source models with CI/CD deployment support and provenance tracking.
Spatio-temporal Explanation for Adversarial-Aware Cloud Vision Services
Zerui Wang, Yan Liu
IEEE Computer Society Signature Conference on Computers, Software, and Applications (COMPSAC)
Extends spatio-temporal XAI capabilities to cloud vision services with adversarial awareness.
The Analysis and Development of an XAI Process on Feature Contribution Explanation
Jiahao Huang*, Zerui Wang*, Duanyu Li, Yan Liu (*Equal contribution)
IEEE International Conference on Big Data
Foundational work formalizing the XAI process and introducing quantitative metrics for evaluating explanation consistency and stability.
A Trustworthy View on Explainable Artificial Intelligence Method Evaluation
Ding Li, Yan Liu, Jun Huang, Zerui Wang
IEEE Computer
Survey paper providing a trustworthy perspective on XAI method evaluation, addressing measurement and evaluation challenges.
Linking Team-level and Organization-level Governance in Machine Learning Operations through Explainable AI and Responsible AI Connector
Elie Neghawi, Zerui Wang, Jun Huang, Yan Liu
IEEE 47th Annual Computers, Software, and Applications Conference (COMPSAC)
Links team-level and organization-level governance in MLOps through XAI and Responsible AI connectors.
VideoXAI - A Hybrid Architecture for Explainable AI Pipelines of Robust Video Classification
Abideep Singh Kondal, Ravinder Singh Ghataura, Yan Liu, Zerui Wang
IEEE International Conference on Big Data
Hybrid architecture combining spatio-temporal feature extraction with multi-modal fusion for robust and explainable video classification.
Other Publications
1 papersThe Role of Provenance Modeling in Tracing and Reproducing Explainable AI Pipelines
Zerui Wang, Yan Liu
World Congress in Computer Science, Computer Engineering, & Applied Computing (CSCE)
Introduces provenance modeling techniques ensuring reproducibility across XAI experiments.
Citation data from Google Scholar. Last updated: January 2025.