Yao-Hung Hubert Tsai

Yao-Hung Hubert Tsai Email: yaohungt [at] cs (dot) cmu (dot) edu [Google Scholar] [DBLP] [Github]

About Me

I obtained my Ph.D. degree in Machine Learning Department at Carnegie Mellon University under the supervision of Dr. Ruslan Salakhutdinov and Dr. Louis-Philippe Morency. I did internships at Facebook AI Research, Apple AI Research, Allen Institute for Artificial Intelligence (AI2), and Microsoft Research. I was a visiting scholar at Kokuritsu Kenkyū Kaihatsu Hōjin Rikagaku Kenkyūjyo Center for Advanced Intelligence Project (RIKEN AIP) and Kyoto University. I have been supported by AI2 and Facebook fellowships. This site is no longer maintained.

Research Summary

My research focuses on building intelligent agents, which leverage massive heterogeneous, noisy and imperfect information to learn from, reason about and interact with the real world.

I care about the following topics (with selected first-authored publications):

• Multimodal Learning: summarize heterogeneous sources of information from multi-modal sensors.
• Multimodal Generation [1], Multimodal Fusion [2] and Multimodal Interpretation [3].
• Machine Reasoning: give human levels of problem solving skills and human capacity in generalization.
• Uncertainty Quantification [1], Relationship Modeling [2, 3] and Common Sense Reasoning [4].
• Transfer Learning: enable the experience sharing across various domains and tasks.
• Domain Adaptation [1], Multitask Learning [2] and Few-shot Learning [3, 4].
• Less-supervised Learning: make use of massive noisy and imperfect information.
• Semi-supervised Learning, Weakly-supervised Learning and Self-supervised Learning [1, 2].