Publications

(2026). Your Compiler is Backdooring Your Model: Understanding and Exploiting Compilation Inconsistency Vulnerabilities in Deep Learning Compilers. In S&P 2026 🏆 Distinguished Paper Award.

PDF Code

(2026). IAG: Input-aware Backdoor Attack on VLM-based Visual Grounding. In CVPR 2026.

(2026). FedRE: A Representation Entanglement Framework for Model-Heterogeneous Federated Learning. In CVPR 2026.

(2026). CodeSense: a Real-World Benchmark and Dataset for Code Semantic Reasoning. In ICLR 2026.

(2026). From Assistant to Independent Developer — Are GPTs Ready for Software Development?. In ICLR 2026.

(2025). PARD: Enhancing Goodput for Inference Pipeline via Proactive Request Dropping. In EuroSys 2026.

(2025). TITLE_TODO -- please fill in the EMNLP 2025 paper title. In EMNLP 2025.

(2025). SOK: Efficiency Robustness of Dynamic Deep Learning Systems. In USENIX Security 2025.

(2025). FDPT: Federated Discrete Prompt Tuning for Black-Box Visual-Language Models. In ICCV 2025.

(2025). Medusa: A Framework for Collaborative Development of Foundation Models with Automated Parameter Ownership Assignment. In FSE 2025.

(2024). DeciX: Explain Deep Learning Based Code Generation Applications. In ESEC/FSE 2024.

(2024). PPM: Automated Generation of Diverse Programming Problems for Benchmarking Code Generation Models. In ESEC/FSE 2024.

PDF Code

(2023). Dynamic Transformer Provide a False Sense of Efficiency. In ACL 2023.

(2023). The Dark Side of Dynamic Routing Neural Networks: Towards Efficiency Backdoor Injection. In CVPR 2023.

(2023). Sibling-Attack: Rethinking Transferable Adversarial Attacks against Face Recognition. In CVPR 2023.

(2023). DyCL: Dynamic Neural Network Compilation Via Program Rewriting and Graph Optimization. In ISSTA 2023.

(2022). NMTSloth: Understanding and Testing Efficiency Degradation of Neural Machine Translation Systems. In ESEC/FSE 2022.

PDF Code

(2022). Learning to Reverse DNNs from AI Programs Automatically. In IJCAI 2022.

PDF

(2022). NICGSlowDown: Evaluating the Efficiency Robustness of Neural Caption Generation Models. In CVPR 2022.

PDF Code

(2020). DENAS: automated rule generation by knowledge extraction from neural networks. In ESEC/FSE 2020.

PDF Code DOI