Hang Su
Associate Professor
Works on trustworthy and explainable AI, adversarial robustness, computer vision, and reinforcement learning.
Hang Su's lab at Tsinghua works on trustworthy & safe ML, reinforcement learning and interpretability, evaluation & safety of LMs. 103 papers since 2023, 73 at top venues, mostly ICML and ICLR. Applications close .
At a glance
- papers, 2023–2026
- 27/ year
- at T1 venues
- 71%
- PhD students
- —
- MS students
- —
- alumni listed
- —
Papers by yearfrom DBLP, 2023–2026
- 15 papers2026 · T1 13 · collecting
- 24 papers2025 · T1 14
- 36 papers2024 · T1 27
- 28 papers2023 · T1 19
T1 venuesOtherVenues: ICML, ICLR, NeurIPS, TPAMI, CVPR
Research focuspaper titles that use each topic's words
- adversarial30 papers
- Reinforcement learning23 papers
- Trustworthy & safe ML2 papers
- Interpretability, evaluation & safety of LMs1 paper
- adversarial robustness0 papers
Paper list
From DBLP.
202615 papers
- TPAMI (T1)A Survey on Autonomy-Induced Security Risks in Large Model-Based Agents (opens in a new tab)
- IJCV (T1)AEMIM: Adversarial Examples Meet Masked Image Modeling (opens in a new tab)
- AAAI (T1)Benchmarking Trustworthiness in Multimodal LLMs for Video Understanding (opens in a new tab)
- IEEE CAA J. Autom. SinicaCrafting Physical Adversarial Examples by Combining Differentiable and Physically Based Renders (opens in a new tab)
- ICLR (T1)DiffusionNFT: Online Diffusion Reinforcement with Forward Process (opens in a new tab)
- AAAI (T1)Dual-Seed Evolutionary Algorithm for Noise Optimization in Diffusion Models (opens in a new tab)
- ICLR (T1)Exploratory Diffusion Model for Unsupervised Reinforcement Learning (opens in a new tab)
- AAAI (T1)H-RDT: Human Manipulation Enhanced Bimanual Robotic Manipulation (opens in a new tab)
- ICLR (T1)Helix: Evolutionary Reinforcement Learning for Open-Ended Scientific Problem Solving (opens in a new tab)
- TPAMI (T1)Improving Viewpoint Robustness for Visual Recognition via Adversarial Training (opens in a new tab)
- Findings of ACLLCO: LLM-based Constraint Optimization for Safer Agentic LLMs in Real-world Tasks (opens in a new tab)
- IJCV (T1)MLA-Trust: Benchmarking Trustworthiness of Multimodal LLM Agents in GUI Environments (opens in a new tab)
202524 papers
- IJCV (T1)A Comprehensive Study on Robustness of Image Classification Models: Benchmarking and Rethinking (opens in a new tab)
- ICML (T1)Accelerating PDE-Constrained Optimization by the Derivative of Neural Operators (opens in a new tab)
- ICCV (T1)AdvDreamer Unveils: Are Vision-Language Models Truly Ready for Real-World 3D Variations? (opens in a new tab)
- IEEE Trans. Dependable Secur. Comput.Adversarial Examples Detection With Enhanced Image Difference Features Based on Local Histogram Equalization (opens in a new tab)
- IEEE Trans. Big DataANF: Crafting Transferable Adversarial Point Clouds via Adversarial Noise Factorization (opens in a new tab)
- Findings of NAACLAutoBreach: Universal and Adaptive Jailbreaking with Efficient Wordplay-Guided Optimization via Multi-LLMs (opens in a new tab)
- Pattern Recognit. Lett.CamoEnv: Transferable and environment-consistent adversarial camouflage in autonomous driving (opens in a new tab)
- TPAMI (T1)Distributionally Location-Aware Transferable Adversarial Patches for Facial Images (opens in a new tab)
- IEEE Trans. Intell. Transp. Syst.Empowering Object Detection: Unleashing the Potential of Decoupled and Interactive Distillation (opens in a new tab)
- Findings of EMNLPExploring the Generalizability of Factual Hallucination Mitigation via Enhancing Precise Knowledge Utilization (opens in a new tab)
- IEEE Trans. Multim.Exploring Transferability of Multimodal Adversarial Samples for Vision-Language Pre-Training Models With Contrastive Learning (opens in a new tab)
- IJCV (T1)Face3DAdv: Exploiting Robust Adversarial 3D Patches on Physical Face Recognition (opens in a new tab)
202436 papers
- TPAMI (T1)A Comprehensive Survey of Continual Learning: Theory, Method and Application (opens in a new tab)
- IEEE Trans. Vis. Comput. Graph.A Unified Interactive Model Evaluation for Classification, Object Detection, and Instance Segmentation in Computer Vision (opens in a new tab)
- NeurIPS (T1)Aligning Diffusion Behaviors with Q-functions for Efficient Continuous Control (opens in a new tab)
- Neural Comput. Appl.Black-box attacks on face recognition via affine-invariant training (opens in a new tab)
- ECCV (T1)Controllable Navigation Instruction Generation with Chain of Thought Prompting (opens in a new tab)
- ECCV (T1)CRM: Single Image to 3D Textured Mesh with Convolutional Reconstruction Model (opens in a new tab)
- ECCV (T1)DIFFender: Diffusion-Based Adversarial Defense Against Patch Attacks (opens in a new tab)
- NeurIPS (T1)Diffusion Models are Certifiably Robust Classifiers (opens in a new tab)
- ICML (T1)DPOT: Auto-Regressive Denoising Operator Transformer for Large-Scale PDE Pre-Training (opens in a new tab)
- IEEE Trans. Intell. Transp. Syst.Efficient Adversarial Attack Strategy Against 3D Object Detection in Autonomous Driving Systems (opens in a new tab)
- ICLR (T1)Embodied Active Defense: Leveraging Recurrent Feedback to Counter Adversarial Patches (opens in a new tab)
- ACM MMExploring the Robustness of Decision-Level Through Adversarial Attacks on LLM-Based Embodied Models (opens in a new tab)
202328 papers
- Comput. Vis. Image Underst.AdvFAS: A robust face anti-spoofing framework against adversarial examples (opens in a new tab)
- CVPR (T1)All are Worth Words: A ViT Backbone for Diffusion Models (opens in a new tab)
- CVPR (T1)Benchmarking Robustness of 3D Object Detection to Common Corruptions in Autonomous Driving (opens in a new tab)
- ICLR (T1)Bi-level Physics-Informed Neural Networks for PDE Constrained Optimization using Broyden's Hypergradients (opens in a new tab)
- ICCV (T1)COCO-O: A Benchmark for Object Detectors under Natural Distribution Shifts (opens in a new tab)
- Pattern Recognit. Lett.Consistent attack: Universal adversarial perturbation on embodied vision navigation (opens in a new tab)
- ICML (T1)Contrastive Energy Prediction for Exact Energy-Guided Diffusion Sampling in Offline Reinforcement Learning (opens in a new tab)
- ICCV (T1)Detection Transformer with Stable Matching (opens in a new tab)
- ICLR (T1)DINO: DETR with Improved DeNoising Anchor Boxes for End-to-End Object Detection (opens in a new tab)
- AAAI (T1)DQ-DETR: Dual Query Detection Transformer for Phrase Extraction and Grounding (opens in a new tab)
- ICML (T1)GNOT: A General Neural Operator Transformer for Operator Learning (opens in a new tab)
- NeurIPS (T1)Hierarchical Decomposition of Prompt-Based Continual Learning: Rethinking Obscured Sub-optimality (opens in a new tab)
Where alumni went
Alumni are not listed on the lab homepage.
Apply to TsinghuaFall 2027
Computer Science PhD (international applicants)
- Round
- Main
- Opens
- GRE
- See program page
- English
- Per school catalog
- Fee
- RMB 800
Not yet confirmed for this intake.
Similar labsnearest by research summary
- USMichigan
Saptarshi Chakraborty
- Learning theory & optimization
- USCMU
Ruslan Salakhutdinov
- Reinforcement learning
- USUIUC
Varun Chandrasekaran
- Trustworthy & safe ML
- SGNTU
Dacheng Tao
- Learning theory & optimization
- HKCUHK
Wing Cheong Lau
- Learning theory & optimization
- SGNTU
Wooi Boon Goh
- Data management & big data
Your notes
Save this lab to keep notes.
Data
Some numbers could not be verified:
- The member list could not be found on the lab site.
- The alumni list could not be found on the lab site.
- Collected
- Verified
- Papers
- DBLP
- Members
- Not collected
- Alumni
- Not collected
Tiers are venue classifications, not lab rankings.
Tsinghuacloses