Su-In Lee
Professor
Develops explainable AI methods for biomedicine, including feature attribution for clinical and genomic models.
Su-In Lee's lab at UW works on medical imaging & healthcare AI, genomics & computational biology and trustworthy & safe ML. 24 papers since 2023, 8 at top venues, mostly ICLR and Nature Machine Intelligence. 8 PhD students. 8 alumni, most often to Microsoft and OpenAI. Applications close .
At a glance
- papers, 2023–2026
- 6/ year
- at T1 venues
- 33%
- PhD students
- 8
- MS students
- 0
- alumni listed
- 8
Papers by yearfrom the lab homepage, 2023–2026
- 3 papers2026 · T1 1 · collecting
- 3 papers2025 · T1 0
- 6 papers2024 · T1 2
- 12 papers2023 · T1 5
T1 venuesOtherVenues: ICLR, Nature Machine Intelligence, ICML, Nature Communications, Nature Methods
Research focuspaper titles that use each topic's words
- feature7 papers
- attribution4 papers
- Medical imaging & healthcare AI3 papers
- Genomics & computational biology2 papers
- Trustworthy & safe ML1 paper
Paper list
From the lab homepage.
20263 papers
- Nature Machine Intelligence (T1)A deep generative model of single-cell methylomic data (opens in a new tab)
- Accepted to ICMLEnsembling Sparse Autoencoders (opens in a new tab)
- Accepted to ICMLSurrogateSHAP: Training-Free Contributor Attribution for Text-to-Image (T2I) Models (opens in a new tab)
20253 papers
- Neural Information Processing Systems (NeurIPS)CellCLIP - Learning Perturbation Effects in Cell Painting via Text-Guided Contrastive Learning (opens in a new tab)
- International Conference on Representation LearningEfficient Shapley Values for Attributing Global Properties of Diffusion Models to Data Group (opens in a new tab)
- Nature Reviews BioengineeringTransparency in medical AI systems across the development pipeline (opens in a new tab)
20246 papers
- Nature Biomedical EngineeringA deep profile of gene expression across 18 human cancers (opens in a new tab)
- Won the best paper runner-up award at the Data Curation and Augmentation in Medical Imaging Workshop at CVPRDiscovering mechanisms underlying AI prediction of protected attributes via data auditing (opens in a new tab)
- ICLR (T1)Estimating Conditional Mutual Information for Dynamic Feature Selection (opens in a new tab)
- NeurIPS (T1)Stochastic Amortization: A Unified Approach to Accelerate Feature and Data Attribution
- The LancetThe clinical potential of counterfactual AI models (opens in a new tab)
- Nature MedicineTransparent medical image AI via an image-text foundational model grounded in medical literature (opens in a new tab)
202312 papers
- Nature Machine Intelligence (T1)Algorithms to estimate Shapley value feature attributions (opens in a new tab)
- Lancet Healthy Longevity - featured on the coverAn explainable AI framework for interpretable biological age (opens in a new tab)
- Nature Biomedical EngineeringAuditing the inference processes of medical-image classifiers by leveraging generative AI and the expertise of physicians (opens in a new tab)
- ICLR (T1)Contrastive Corpus Attribution for Explaining Representations (opens in a new tab)
- Neural Information Processing Systems (NeurIPS)Feature selection in the contrastive analysis setting
- Nature Methods (T1)Isolating salient variations of interest in single-cell transcriptomic data with contrastiveVI (opens in a new tab)
- International Conference on Learning Representations (ICLR) spotlightLearning to estimate Shapley values with vision transformers (opens in a new tab)
- ICML (T1)Learning to Maximize Mutual Information for Dynamic Feature Selection (opens in a new tab)
- Neural Information Processing Systems (NeurIPS)On the robustness of removal-based feature attributions (opens in a new tab)
- Genome Biology 24, Article numberPAUSE: principled feature attribution for unsupervised gene expression analysis (opens in a new tab)
- Nature Communications (T1)PERSIST: Predictive and robust gene selection for spatial transcriptomics (opens in a new tab)
- Nature Biomedical Engineering 7, 811-829Uncovering expression signatures of synergistic drug responses via ensembles of explainable machine-learning models (opens in a new tab)
Where alumni went8 listed, 67% to industry
Industry 4Academia 2
Most often to
- Microsoft
- OpenAI
- Uber
Also to
- Harvard University
- Stanford University
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
- GBOxford
Anshul Thakur
- Medical imaging & healthcare AI
- GBCambridge
Mihaela van der Schaar
- Medical imaging & healthcare AI
- CAToronto
Chris McIntosh
- Medical imaging & healthcare AI
- USUC Berkeley
Guanghan Meng
- Neuroscience & brain-computer interfaces
- USNYU
Laure Zanna
- Physics, climate & scientific computing
- USGeorgia Tech
May Dongmei Wang
- Medical imaging & healthcare AI
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