Zhangyang Wang
Associate Professor
Leads the VITA Group, working on efficient and robust deep learning, sparse neural networks and large language models.
Zhangyang Wang's lab at UT Austin works on trustworthy & safe ML, large language models and efficient training & inference. 197 papers since 2023, 165 at top venues, mostly ICLR and ICML. 15 PhD students. 45 alumni, most often to Amazon and Apple. Applications close .
At a glance
- papers, 2023–2026
- 51/ year
- at T1 venues
- 84%
- PhD students
- 15
- MS students
- 0
- alumni listed
- 45
Papers by yearfrom the lab homepage, 2023–2026
- 39 papers2026 · T1 27 · collecting
- 44 papers2025 · T1 32
- 52 papers2024 · T1 48
- 62 papers2023 · T1 58
T1 venuesOtherVenues: ICLR, ICML, NeurIPS, CVPR, ICCV
Research focuspaper titles that use each topic's words
- Large language models69 papers
- Efficient training & inference32 papers
- sparse7 papers
- Trustworthy & safe ML4 papers
- robust deep0 papers
Paper list
From the lab homepage.
202639 papers
- ECCV (T1)A Physics-Grounded Benchmark for Multi-Agent Dynamics in World Models
- SPMAn Overview of Low-Rank Structures in the Training and Adaptation of Large Models
- ICML (T1)Beyond Test-Time Memory: State-Space Optimal Control for LLM Reasoning
- COLMCan Induced Emotion Bias LLM Behaviors in Sequential Decision Making?
- COLMCorrect Answers from Sound Reasoning: Verifiable Process Supervision for Language Models
- Advances in Neural Information Processing Systems (NeurIPS), Evaluations and Datasets TrackDD-Ranking: Rethinking the Evaluation of Dataset Distillation
- ICML (T1)Fantastic Reasoning Behaviors and Where to Find Them: Unsupervised Discovery of the Reasoning Process
- ICAIFFrom Hindsight to Foresight: Adapting Reusable Reasoning Structures from Recurring Financial Events
- NeurIPS (T1)From Link Prediction to Linear Community Detection: A Finite-Sample Guarantee for Graph Pretraining
- ICML (T1)GradientStabilizer: Fix the Norm, Not the Gradient
- CVPR (T1)HumanNOVA: Photorealistic, Universal and Rapid 3D Human Avatar Modeling from a Single Image
- ICRA (T1)LAD-VF: LLM-Automatic Differentiation Enables Fine-Tuning-Free Robot Planning from Formal Methods Feedback
202544 papers
- ICLR (T1)4K4DGen: Panoramic 4D Generation at 4K Resolution
- CVPR (T1)A Stitch in Time Saves Nine: Small VLM is a Precise Guidance for accelerating Large VLMs
- MLSys (T1)APOLLO: SGD-like Memory AdamW-level Performance
- TMLR (T1)AutoTrust: Benchmarking Trustworthiness in Large Vision Language Models for Autonomous Driving
- COLMCan Scaling Test-Time Compute Improve World Foundation Model?
- ICML (T1)Cavia: Camera-controllable Multi-view Video Diffusion with View-Integrated Attention
- JAMAComplete AI-Enabled Echocardiography Interpretation With Multitask Deep Learning
- DMLRDeep Learning for Accurate Diagnosis of Viral Infections through scRNA-seq Analysis: A Comprehensive Benchmark Study
- NeurIPS (T1)Drag-and-Drop LLMs: Zero-Shot Prompt-to-Weights
- CVPR (T1)Feature4X: Bridging Any Monocular Video to 4D Agentic AI with Versatile Gaussian Feature Fields
- CVPR (T1)FlexGS: Train Once Deploy Everywhere with Many-in-One Flexible 3D Gaussian Splatting
- ICML (T1)From Low Rank Gradient Subspace Stabilization to Low-Rank Weights: Observations, Theories and Applications
202452 papers
- JAMA CardiologyA Multimodality Video-Based AI Biomarker for Aortic Stenosis Development and Progression
- TMLR (T1)AGG: Amortized Generative 3D Gaussians for Single Image to 3D
- NeurIPS (T1)AlphaPruning: Using Heavy-Tailed Self Regularization Theory for Improved Layer-wise Pruning of Large Language Models
- Lancet Digital HealthArtificial intelligence-guided detection of under-recognized cardiomyopathies on point-of-care cardiac ultrasound: a multi-center study
- ICLR (T1)Compressing LLMs: The Truth is Rarely Pure and Never Simple
- ICLR (T1)Data Distillation Can Be Like Vodka: Distilling More Times For Better Quality
- ICML (T1)Decoding Compressed Trust: Scrutinizing the Trustworthiness of Efficient LLMs Under Compression
- NeurIPS (T1)Diffusion4D: Fast Spatial-temporal Consistent 4D generation via Video Diffusion Models
- ICLR (T1)Doubly Robust Instance-Reweighted Adversarial Training
- ICLR (T1)DP-OPT: Make Large Language Model Your Differentially-Private Prompt Engineer
- ECCV (T1)DreamScene360: Unconstrained Text-to-3D Scene Generation with Panoramic Gaussian Splatting
- ICLR (T1)Efficient-3DiM: Learning a Generalizable Single-image Novel-view Synthesizer in One Day
202362 papers
- TIP (T1)A Multi-Purpose Real Haze Benchmark with Quantifiable Haze Levels and Ground Truth
- ICCV (T1)AdaMV-MoE: Adaptive Multi-Task Vision Mixture-of-Experts
- CVPR (T1)AligNeRF: High-Fidelity Neural Radiance Fields via Alignment-Aware Training
- ICML (T1)Are Large Kernels Better Teachers than Transformers for ConvNets?
- TMLR (T1)Can Pruning Improve Certified Robustness of Neural Networks?
- TMLR (T1)Chasing Better Deep Image Priors between Over- and Under-Parameterization
- ICLR (T1)Continuous-Discrete Convolution for (3+1)D Geometry-Sequence Modeling in Proteins
- TMLR (T1)CR-MoE: Consistent Routed Mixture-of-Experts for Scaling Contrastive Learning
- ICML (T1)Data Efficient Neural Scaling Law via Model Reusing
- NeurIPS (T1)Don't Just Prune by Magnitude! Your Mask Topology is A Secret Weapon
- ACL (T1)DSEE: Dually Sparsity-embedded Efficient Tuning of Pre-trained Language Models
- NeurIPS (T1)Dynamic Sparsity Is Channel-Level Sparsity Learner
Where alumni went45 listed, 76% to industry
Industry 32Academia 10
Most often to
- Amazon
- Apple
- NVIDIA
Also to
- Cornell University
- CUHK
- MIT
Apply to UT AustinFall 2027
Computer Science PhD
- Round
- Single round (Dec 15)
- Opens
- GRE
- See program page
- English
- See program page
- Fee
- See program page
Not yet confirmed for this intake.
Similar labsnearest by research summary
- CNSJTU
Zheng Huang
- Trustworthy & safe ML
- CNTsinghua
Jun Zhu
- Learning theory & optimization
- CNSJTU
Junchi Yan
- Learning theory & optimization
- GBOxford
Varun Kanade
- Learning theory & optimization
- CNZhejiang
Kui Ren
- Trustworthy & safe ML
- USCMU
J. Zico Kolter
- 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.
UT Austincloses