Matt Y. Cheung
I am a PhD student in Electrical and Computer Engineering at Rice University advised by Ashok Veeraraghavan and Guha Balakrishnan. I work on uncertainty quantification, specifically Conformal Prediction for medical imaging problems, as well as generative models and hallucinations. I was a trainee in the NIH NLM Training Program in Biomedical Informatics and Data Science. I received my M.S. in Electrical and Computer Engineering from Rice University in 2023 and my B.S. in Electrical Engineering (cum laude) from UC Davis in 2020.
📢 I am looking for full-time opportunities in the Bay Area CA!
news
| Aug 10, 2026 | Our work at Johnson & Johnson R&D “GAIZ: Automated Detection of Geographic Atrophy Biomarkers via Volumetric OCT using Novel Morphological Objectives” was accepted at the 2026 MLMI Workshop at MICCAI as an Oral. |
|---|---|
| Jun 12, 2026 | Our paper Efficient Conformal Volumetry for Template-Based Segmentation was accepted at MICCAI 2026 as a Spotlight. |
| May 29, 2026 | New paper on arXiv Conformal Certification of Reasoning Trace Prefixes. |
| May 11, 2026 | I am working at Johnson & Johnson R&D this summer as a AI/ML Computer Vision intern. |
| Jan 26, 2026 | Our paper COMPASS: Robust Feature Conformal Prediction for Medical Segmentation Metrics was accepted at ICLR 2026. |
latest posts
| Jul 23, 2026 | Notes on Transformer Positional Encodings |
|---|---|
| Jul 22, 2026 | Notes on Transformer Blocks, Tensor Dimensions and Attention FLOPs |
| Jul 18, 2026 | (Running) Notes on Language Modeling from Scratch |
| Jun 18, 2026 | RL Notes: TRPO and PPO |
| Mar 28, 2026 | Survey of Conformal Predictions for LLMs |
selected publications
-
COMPASS: Robust Feature Conformal Prediction for Medical Segmentation MetricsICLR, 2026 -
Metric-Guided Conformal Bounds for Probabilistic Image ReconstructionUNSURE Workshop at MICCAI (Long Oral), 2025 -
Wearable blood pressure monitoring devices: Understanding heterogeneity in design and evaluationIEEE Transactions on Biomedical Engineering, 2024