Welcome to my academic homepage. I am a Ph.D. candidate in
Computer Science at University of California, Los Angeles (UCLA)
I work on representation learning for visual reasoning and skill learning tasks. In particular, I'm interested in building and understanding inductive biases for learning representations from multi-modal data, so as to systematically generalize in real-world physical and psychological dynamics. Some of my research keywords can be found below:
Humanlike Learning: Concept and language acqusition, Embodied agents, (Inverse) Reinforcement learning
Representation Learning: OOD generalization (domain and systematic), Energy-based generative model, Multi-modal learning
[06/2022] Will work as a research intern at .
[05/2022] One paper on reconciling diffusion model and latent-space enery-based model is accepted to . ICML 2022 Code has been released.
[02/2022] One paper on benchmarking few-shot compositional visual reasoning with real-world data is accepted as oral to . CVPR 2022 Code has been released.
[01/2022] One paper on human-like learning for systematic generalization in visual relational reasoning is accepted to . ICLR 2022 Code has been released.
2021.6 - 2021.12, Research Intern, NVIDIA Research
Mentors: Dr. Weili Nie, Dr. Yuke Zhu and Dr. Anima Anandkumar.
2020.6 - 2021.5, Research Intern, Google Brain Robotics
Mentors: Dr. Pannag Sanketi and Laura Graesser.
2019.6 - 2019.9, Research Intern, ByteDance AI Lab
Mentors: Dr. Tao Kong and Dr. Lei Li.
2017.7 - 2017.9, Research Intern, CVRP Lab, School of Computer Science, National University of Singapore
Mentor: Dr. Gim Hee Lee.
Conference PC Member/Reviewer: ICML (top reviewer), NeurIPS, ICLR, AAAI, IJCAI (senior PC), CVPR (outstanding reviewer), ICCV, ECCV, WACV, ACCV, RSS, ICRA, IROS, CoRL
Journal Reviewer: TMLR, IEEE RA-L, AAS, IEEE/CAA JAS
UCLA Center for Vision, Cognition, Learning and Autonomy
603 Charles E Young Dr E #3878
Los Angeles, CA 90024
xm [at] cs [dot] ucla [dot] edu
[ G o o g l e Scholar]