J. Zico Kolter
Professor and Department Head
Studies robust and reliable machine learning, including LLM safety and security, optimization and implicit deep models.
J. Zico Kolter's lab at CMU works on trustworthy & safe ML and learning theory & optimization. 30 papers since 2023, 27 at top venues, mostly ICLR and NeurIPS. 16 alumni, most often to Google and Meta. Applications close .
At a glance
- papers, 2023–2026
- 8/ year
- at T1 venues
- 90%
- PhD students
- 0
- MS students
- 0
- alumni listed
- 16
Papers by yearfrom the lab homepage, 2023–2026
- 0 papers2026 · T1 0 · collecting
- 0 papers2025 · T1 0
- 11 papers2024 · T1 11
- 19 papers2023 · T1 16
T1 venuesOtherVenues: ICLR, NeurIPS, CVPR, ICML, AAAI
Research focuspaper titles that use each topic's words
- Learning theory & optimization7 papers
- Trustworthy & safe ML1 paper
- implicit1 paper
- studies robust0 papers
- reliable machine0 papers
Paper list
From the lab homepage.
202411 papers
- ICLR (T1)A simple and effective pruning approach for large language models (opens in a new tab)
- UAI (T1)Computing Low-Entropy Couplings for Large-Support Distributions (opens in a new tab)
- CVPR (T1)DART: Implicit Doppler Tomography for Radar Novel View Synthesis (opens in a new tab)
- ICLR (T1)Manifold preserving guided diffusion (opens in a new tab)
- ICLR (T1)On the Joint Interaction of Models, Data, and Features (opens in a new tab)
- CVPR (T1)Scaling Laws for Data Filtering–Data Curation cannot be Compute Agnostic (opens in a new tab)
- ICLR (T1)T-mars: Improving visual representations by circumventing text feature learning (opens in a new tab)
- ICLR (T1)The Update Equivalence Framework for Decision-Time Planning (opens in a new tab)
- ICLR (T1)Understanding Augmentation-based Self-Supervised Representation Learning via RKHS Approximation (opens in a new tab)
- ICLR (T1)Understanding prompt engineering may not require rethinking generalization (opens in a new tab)
- ICLR (T1)Why is SAM Robust to Label Noise? (opens in a new tab)
202319 papers
- ICML (T1)Abstracting imperfect information away from two-player zero-sum games (opens in a new tab)
- WACVAdversarial robustness in discontinuous spaces via alternating sampling \& descent (opens in a new tab)
- ICML (T1)Can neural network memorization be localized? (opens in a new tab)
- NeurIPS (T1)Deep Equilibrium Based Neural Operators for Steady-State PDEs (opens in a new tab)
- —Deep Off-Policy Iterative Learning Control (opens in a new tab)
- CVPR (T1)Finetune like you pretrain: Improved finetuning of zero-shot vision models (opens in a new tab)
- AAAI (T1)Function approximation for solving stackelberg equilibrium in large perfect information games (opens in a new tab)
- NeurIPS (T1)Language models are weak learners (opens in a new tab)
- NeurIPS (T1)Learning with Explanation Constraints (opens in a new tab)
- AAAI (T1)Losses over labels: Weakly supervised learning via direct loss construction (opens in a new tab)
- ICML (T1)Mimetic initialization of self-attention layers (opens in a new tab)
- EMNLP (T1)Model-tuning Via Prompts Makes NLP Models Adversarially Robust (opens in a new tab)
Where alumni went16 listed, 73% to industry
Industry 11Academia 4
Most often to
- Meta
- Bosch
Also to
- MIT
- NUS
- UIUC
Apply to CMUFall 2027
CMU SCS Computer Science PhD
- Round
- Single round
- Opens
- GRE
- See program page
- English
- TOEFL / IELTS / Duolingo (F-1/J-1 non-native speakers)
- Fee
- USD 100 (USD 80 by early deadline)
Confirmed on the program page.
Similar labsnearest by research summary
- USStanford
Sanjay Lall
- Learning theory & optimization
- USCMU
Aditi Raghunathan
- Trustworthy & safe ML
- CNSJTU
Junchi Yan
- Learning theory & optimization
- GBOxford
Varun Kanade
- Learning theory & optimization
- USNYU
Pavel Izmailov
- Trustworthy & safe ML
- USUT Austin
Zhangyang Wang
- 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.
CMUcloses