Sewoong Oh
Professor
Researches machine learning theory and privacy, including differential privacy and tokenizers for language models.
Sewoong Oh's lab at UW works on learning theory & optimization, trustworthy & safe ML and large language models. 53 papers since 2023, 39 at top venues, mostly ICML and NeurIPS. 13 alumni, most often to Amazon and Anthropic. Applications close .
At a glance
- papers, 2023–2026
- 14/ year
- at T1 venues
- 74%
- PhD students
- 0
- MS students
- 0
- alumni listed
- 13
Papers by yearfrom the lab homepage, 2023–2026
- 12 papers2026 · T1 7 · collecting
- 8 papers2025 · T1 6
- 16 papers2024 · T1 13
- 17 papers2023 · T1 13
T1 venuesOtherVenues: ICML, NeurIPS, CVPR, ICLR, CCS
Research focuspaper titles that use each topic's words
- Large language models17 papers
- Learning theory & optimization12 papers
- privacy7 papers
- differential3 papers
- Trustworthy & safe ML0 papers
Paper list
From the lab homepage.
202612 papers
- SatMLA False Sense of Privacy: Evaluating Textual Data Sanitization Beyond Surface-level Privacy Leakage (opens in a new tab)
- ICML (T1)Anchored Decoding: Provably Reducing Copyright Risk for Any Language Model (opens in a new tab)
- SatMLAre Robust LLM Fingerprints Adversarially Robust? (opens in a new tab)
- CVPR (T1)Concept-Aware Batch Sampling Improves Language-Image Pretraining (opens in a new tab)
- COLMCorrect Answers from Sound Reasoning: Verifiable Process Supervision for Language Models (opens in a new tab)
- ICLR (T1)OpenThoughts: Data Recipes for Reasoning Models (opens in a new tab)
- Molecular & Cellular ProteomicsPretrained spectrum representations for diverse tandem mass spectrometry proteomics tasks (opens in a new tab)
- ICML (T1)Privasis: Synthesizing the Largest “Public” Private Dataset from Scratch (opens in a new tab)
- ICML (T1)Sampling from Your Language Model One Byte at a Time (opens in a new tab)
- ICML (T1)Spurious Rewards: Rethinking Training Signals in RLVR (opens in a new tab)
- Annals of Applied ProbabilityStochastic optimization on matrices and a graphon McKean–Vlasov limit (opens in a new tab)
- ICML (T1)Zeroth-Order Optimization at the Edge of Stability (opens in a new tab)
20258 papers
- ICML (T1)Economics of Sourcing Human Data (opens in a new tab)
- CVPR (T1)PLeaS–Merging Models with Permutations and Least Squares (opens in a new tab)
- COLMRecycling the Web: A Method to Enhance Pre-training Data Quality and Quantity for Language Models (opens in a new tab)
- ICML (T1)S4S: Solving for a Diffusion Model Solver (opens in a new tab)
- NeurIPS (T1)Scalable Fingerprinting of Large Language Models (opens in a new tab)
- COLMSuperBPE: Space Travel for Language Models (opens in a new tab)
- NeurIPS (T1)Understanding the Gain from Data Filtering in Multimodal Contrastive Learning (opens in a new tab)
- NeurIPS (T1)Zeroth-Order Optimization Finds Flat Minima (opens in a new tab)
202416 papers
- Journal of Proteome ResearchAccounting for digestion enzyme bias in Casanovo (opens in a new tab)
- Harvard Data Science Review, 6(1), JanuaryAdvancing Differential Privacy: Where We Are Now and Future Directions for Real-World Deployment (opens in a new tab)
- CCS (T1)Air Gap: Protecting Privacy-Conscious Conversational Agents (opens in a new tab)
- EMNLP (T1)Better Alignment with Instruction Back-and-Forth Translation (opens in a new tab)
- NAACL (T1)Can Public Large Language Models Help Private Cross-device Federated Learning? (opens in a new tab)
- NeurIPS (T1)Data Mixture Inference: What do BPE Tokenizers Reveal about their Training Data? (opens in a new tab)
- NeurIPS (T1)DataComp-LM: In search of the next generation of training sets for language models DataComp-LM: In search of the next generation of training sets for language models (opens in a new tab)
- ICML (T1)DeepPolar: Inventing Nonlinear Large-Kernel Polar Codes via Deep Learning (opens in a new tab)
- ICML (T1)DPZero: Private Fine-Tuning of Language Models without Backpropagation (opens in a new tab)
- ICML (T1)Improved Communication-Privacy Trade-offs in Mean Estimation under Streaming Differential Privacy (opens in a new tab)
- COLT (T1)Insufficient Statistics Perturbation: Stable Estimators for Private Least Squares (opens in a new tab)
- NeurIPS (T1)Multilingual Diversity Improves Vision-Language Representations (opens in a new tab)
202317 papers
- ICML (T1)CRISP: Curriculum based Sequential Neural Decoders for Polar Code Family (opens in a new tab)
- NeurIPS (T1)DataComp: In search of the next generation of multimodal datasets (opens in a new tab)
- Journal of Proteome Research, , 22 (11), 3427-3438Evaluating proteomics imputation methods with improved criteria (opens in a new tab)
- ICLR (T1)Few-shot Backdoor Attacks via Neural Tangent Kernels (opens in a new tab)
- Journal of Theoretical ProbabilityGradient flows on graphons: existence, convergence, continuity equations (opens in a new tab)
- NeurIPS (T1)Improving multimodal datasets with image captioning (opens in a new tab)
- NeurIPS (T1)Label Poisoning is All You Need (opens in a new tab)
- CVPR (T1)Learning to Generate Image Embeddings with User-level Differential Privacy (opens in a new tab)
- IEEE Transactions on Selected Areas in Information Theory (JSAIT)Machine Learning-Aided Efficient Decoding of Reed-Muller Subcodes (opens in a new tab)
- JMLR (T1)MAUVE Scores for Generative Models: Theory and Practice (opens in a new tab)
- NeurIPS (T1)Near Optimal Private and Robust Linear Regression (opens in a new tab)
- NeurIPS (T1)On the Connection between Pre-training Data Diversity and Fine-tuning Robustness (opens in a new tab)
Where alumni went13 listed, 67% to industry
Industry 8Academia 4
Most often to
- Amazon
- Anthropic
- ByteDance
Also to
- Shanghai Jiao Tong University
- University of Wisconsin at Madison
- UT Austin
Apply to UWFall 2027
UW Allen School Computer Science PhD
- Round
- Single round
- Opens
- GRE
- not required
- English
- See program page
- Fee
- See program page
Not yet confirmed for this intake.
Similar labsnearest by research summary
- USUMD
Yizheng Chen
- Trustworthy & safe ML
- USMichigan
Clayton Scott
- Learning theory & optimization
- HKCUHK
Wing Cheong Lau
- Learning theory & optimization
- USUIUC
Varun Chandrasekaran
- Trustworthy & safe ML
- USUT Austin
Purnamrita Sarkar
- Learning theory & optimization
- SGNTU
Yang Liu
- 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.
UWcloses