Huan Zhang
Assistant Professor
Studies trustworthy machine learning, including formal verification of neural networks and AI safety and robustness.
Huan Zhang's lab at UIUC works on trustworthy & safe ML, interpretability, evaluation & safety of LMs and compilers & programming languages. 37 papers since 2023, 34 at top venues, mostly NeurIPS and ICLR. 6 PhD students and 8 master's students. 4 alumni. Applications close .
At a glance
- papers, 2023–2026
- 10/ year
- at T1 venues
- 92%
- PhD students
- 6
- MS students
- 8
- alumni listed
- 4
Papers by yearfrom the lab homepage, 2023–2026
- 4 papers2026 · T1 4 · collecting
- 20 papers2025 · T1 17
- 8 papers2024 · T1 8
- 5 papers2023 · T1 5
T1 venuesOtherVenues: NeurIPS, ICLR, ICML, CVPR, EMNLP
Research focuspaper titles that use each topic's words
- Trustworthy & safe ML2 papers
- Compilers & programming languages1 paper
- Interpretability, evaluation & safety of LMs0 papers
- formal verification0 papers
- studies trustworthy0 papers
Paper list
From the lab homepage.
20264 papers
- ICLR (T1)BEAT: Visual Backdoor Attacks on VLM-based Embodied Agents via Contrastive Trigger Learning
- ICLR (T1)DR-SAC: Distributionally Robust Soft Actor-Critic for Reinforcement Learning under Uncertainty
- ICLR (T1)On The Fragility of Benchmark Contamination Detection in Reasoning Models
- ICLR (T1)TokUR: Token-Level Uncertainty Estimation for Large Language Model Reasoning
202520 papers
- NeurIPS (T1)Abstract Rendering: Certified Rendering Under 3D Semantic Uncertainty
- EMNLP (T1)AlphaOne: Reasoning Models Thinking Slow and Fast at Test Time
- ICLR (T1)BaB-ND: Long-Horizon Motion Planning with Branch-and-Bound and Neural Dynamics
- CVPR (T1)Causal Composition Diffusion Model for Closed-loop Traffic Generation
- NeurIPS (T1)Clip-and-Verify: Linear Constraint-Driven Domain Clipping for Accelerating Neural Network Verification
- RSS (T1)CodeDiffuser: Attention-Enhanced Diffusion Policy via VLM-Generated Code for Instruction Ambiguity
- ICLR (T1)DynaMath: A Dynamic Visual Benchmark for Evaluating Mathematical Reasoning Robustness of Vision Language Models
- ICML (T1)EmbodiedBench: Comprehensive Benchmarking Multi-modal Large Language Models for Vision-Driven Embodied Agents
- NeurIPS (T1)GUI-Actor: Coordinate-Free Visual Grounding for GUI Agents
- NeurIPS (T1)Improving Data Efficiency for LLM Reinforcement Fine-tuning Through Difficulty-targeted Online Data Selection and Rollout Replay
- L4DCNeural Contraction Metrics with Formal Guarantees for Discrete-Time Nonlinear Dynamical Systems
- TACASNeural Network Verification with Branch-and-Bound for General Nonlinearities
20248 papers
- ICML (T1)COLD-Attack: Jailbreaking LLMs with Stealthiness and Controllability
- ICML (T1)Fine-grained Local Sensitivity Analysis of Standard Dot-Product Self-Attention
- ICML (T1)Lyapunov-stable Neural Control for State and Output Feedback: A Novel Formulation for Efficient Synthesis and Verification
- NeurIPS (T1)NN4SysBench: Characterizing Neural Network Verification for Computer Systems
- NeurIPS (T1)Regularizing Hidden States Enables Learning Generalizable Reward Model for LLMs
- NeurIPS (T1)Scalable Neural Network Verification with Branch-and-bound Inferred Cutting Planes
- ICML (T1)TrustLLM: Trustworthiness in Large Language Models
- NeurIPS (T1)Verified Safe Reinforcement Learning for Neural Network Dynamic Models
20235 papers
- ICLR (T1)Can Agents Run Relay Race with Strangers? Generalization of RL to Out-of-Distribution Trajectories
- USENIX Security (T1)DiffSmooth: Certifiably Robust Learning via Diffusion Models and Local Smoothing
- ICLR (T1)On the Robustness of Safe Reinforcement Learning under Observational Perturbations
- NeurIPS (T1)Provably Bounding Neural Network Preimages
- ICCV (T1)Robust Mixture-of-Expert Training for Convolutional Neural Networks
Where alumni went4 listed
Apply to UIUCFall 2027
UIUC Siebel Computer Science PhD
- Round
- Single round
- Opens
- —
- GRE
- optional
- English
- TOEFL iBT / IELTS
- Fee
- USD 90
Not yet confirmed for this intake.
Similar labsnearest by research summary
- USUCSD
Berk Ustun
- Learning theory & optimization
- CNSJTU
Yang Yang
- Learning theory & optimization
- USMIT
Sendhil Mullainathan
- Human-AI collaboration & trust
- USNYU
Jaume Vives-i-Bastida
- Reasoning & neuro-symbolic AI
- CAToronto
Xujie Si
- Reasoning & neuro-symbolic AI
- GBImperial
Matthew Wicker
- Trustworthy & safe ML
Your notes
Save this lab to keep notes.
Data
Checked against the lab homepage.
- Collected
- Verified
- Papers
- The lab homepage
- Members
- Lab homepage
- Alumni
- Lab homepage
Tiers are venue classifications, not lab rankings.
UIUCcloses