Hi — I'm Joy, a Ph.D. student in computer science and Knight Hennessy scholar at Stanford University, studying artificial intelligence and computer vision. I am advised by Prof. Jiajun Wu in the CogAI group & Stanford Vision and Learning Lab. My research is graciously funded by Knight Hennessy and the NSF Graduate Research Fellowship.

I finished my B.S. with honors and M.S. with distinction in research at Stanford in 2021, where I was fortunate to be awarded the Ben Wegbreit Prize for best thesis in computer science and the university's Firestone Medal for excellence in research. I was advised by the wonderful Prof. Serena Yeung and Prof. Wah Chiu, and conducted research jointly at Stanford AI Lab and SLAC National Accelerator Laboratory.


My research interests are in visual reasoning and neuro-symbolic learning in the computer vision domain. I’m particularly interested in creating visual understanding models that are generalists, which leverage decomposition and abstraction to achieve reasoning as humans do.

You can reach me at joycj[at]stanford.edu!



Naturally Supervised 3D Visual Grounding with Language-Regularized Concept Learners

Conference on Computer Vision and Pattern Recognition (CVPR) 2024

Chun Feng*, Joy Hsu*, Weiyu Liu, and Jiajun Wu
[paper] [project page]

Learning Planning Abstractions from Language

International Conference on Learning Representations (ICLR) 2024

Weiyu Liu*, Geng Chen*, Joy Hsu, Jiayuan Mao†, and Jiajun Wu†
[paper] [project page]


What’s Left? Concept Grounding with Logic-Enhanced Foundation Models

Conference on Neural Information Processing Systems (NeurIPS) 2023

Joy Hsu*, Jiayuan Mao*, Joshua B. Tenenbaum, and Jiajun Wu
[paper] [project page]

Can Visual Scratchpads With Diagrammatic Abstractions Augment LLM Reasoning?

NeurIPS ICBINB Workshop 2023 [Best Poster Award]

Joy Hsu, Gabriel Poesia, Jiajun Wu, and Noah D. Goodman

Composable Part-Based Manipulation

Conference on Robot Learning (CoRL) 2023

Weiyu Liu, Jiayuan Mao, Joy Hsu, Tucker Hermans, Animesh Garg, and Jiajun Wu
[paper] [project page]

Motion Question Answering via Modular Motion Programs

International Conference on Machine Learning (ICML) 2023

Mark Endo*, Joy Hsu*, Jiaman Li, and Jiajun Wu
[paper] [project page]

NS3D: Neuro-Symbolic Grounding of 3D Objects and Relations

Conference on Computer Vision and Pattern Recognition (CVPR) 2023
CVPR Workshop On Compositional 3D Vision 2023 [Oral Presentation]

Joy Hsu, Jiayuan Mao, and Jiajun Wu
[paper] [project page]

Programmatically Grounded, Compositionally Generalizable Robotic Manipulation

International Conference on Learning Representations (ICLR) 2023 [Notable Top 25%]

Renhao Wang*, Jiayuan Mao*, Joy Hsu, Hang Zhao, Jiajun Wu, and Yang Gao
[paper] [project page]


DisCo: Improving Compositional Generalization in Visual Reasoning through Distribution Coverage

Transactions on Machine Learning Research (TMLR)

Joy Hsu, Jiayuan Mao, and Jiajun Wu
[paper] [project page]

Geoclidean: Few-Shot Generalization in Euclidean Geometry

Conference on Neural Information Processing Systems Datasets and Benchmarks (NeurIPS) 2022

Joy Hsu, Jiajun Wu, and Noah D. Goodman
[paper] [project page]


Unsupervised Learning for Discovery in 2D & 3D Scenes: Towards Unbiased Understanding of Biomedical Images

Stanford CS Honors Thesis [Ben Wegbreit Prize for Best Thesis]

Joy Hsu

Capturing Implicit Hierarchical Structure in 3D Biomedical Images with Self-Supervised Hyperbolic Representations

Conference on Neural Information Processing Systems (NeurIPS) 2021

Joy Hsu*, Jeff Gu*, Gong-Her Wu, Wah Chiu, and Serena Yeung
[paper] [project page]

DARCNN: Domain Adaptive Region-based Convolutional Neural Network for Unsupervised Instance Segmentation in Biomedical Images

Conference on Computer Vision and Pattern Recognition (CVPR) 2021

Joy Hsu, Wah Chiu, and Serena Yeung
[paper] [project page]


Learning Hyperbolic Representations for Unsupervised 3D Segmentation

NeurIPS Differential Geometry Workshop 2020 [Contributed Talk]

Joy Hsu*, Jeff Gu*, and Serena Yeung

Improving Medical Annotation Quality to Decrease Labeling Burden Using Stratified Noisy Cross-Validation

Investigative Ophthalmology & Visual Science, 61(7), 4537-4537.
ACM CHIL Workshop 2020 [Spotlight Talk]

Joy Hsu*, Sonia Phene*, Akinori Mitani, Jieying Luo, Naama Hammel, Jonathan Krause, Rory Sayres


CS 271: Artificial Intelligence in Healthcare [2019, 2020]

Head teaching assistant. CS 271 conducts a deep dive into recent advances in AI in healthcare, focusing in particular on deep learning approaches for healthcare problems. The course covers foundations of neural networks, to cutting-edge deep learning models in the context of image, text, multimodal and time-series data. CS 271 also includes advanced topics on open challenges of integrating AI in a societal application such as healthcare, including interpretability, robustness, privacy and fairness.

CS 41: The Python Programming Language [2018, 2019]

Head teaching assistant. CS 41 teaches the fundamentals and contemporary usage of the Python programming language. The course primarily focuses on developing best practices in writing Python and exploring the extensible and unique parts of Python that make it a powerful language.


  • [Summer 2021] DeepMind research engineering intern

  • [Summer & Fall 2019] Google Medical Brain student researcher

  • [Spring 2018] Apple Photos Search software engineering intern