Scholar
Michael Luo
Google Scholar ID: XpO6-kEAAAAJ
University of California: Berkeley
Citations & Impact
Citations
1,176
H-index
14
i10-index
17
Publications
20
Co-authors
0
Publications
Resume (English only)
Academic Achievements
- - Balsa: Learning a Query Optimizer Without Expert Demonstrations, SIGMOD 2022
- - MESA: Offline Meta-RL for Safe Adaptation and Fault Tolerance, NeurIPS Safe Control Workshop 2021
- - IMPACT: Importance Weighted Asynchronous Architectures with Clipped Target Networks, ICLR 2020
- - Connecting Context-specific Adaptation in Humans to Meta-learning, Preprint
- - Recovery RL: Safe Reinforcement Learning with Learned Recovery Zones, ICRA 2021
- - Distributed Reinforcement Learning is a Dataflow Problem, NeurIPS 2021
Research Experience
- Current research involves building scalable systems for ML practitioners that will fulfill the Sky Computing vision. This includes virtualizing GPUs to scale DL training to trillions of parameters and designing learnable scheduling policies for migrating jobs across different clouds (including on-premise). Master's and undergraduate research primarily focused on practical problems and applications for reinforcement learning (RL), including NLP, query optimization for databases, and video streaming.
Education
- PhD, UC Berkeley EECS, advised by Prof. Ion Stoica; M.S. in EECS under Ion Stoica and Ken Goldberg, graduated in 2021; B.S. from UC Berkeley, double major in EECS and Business Administration, graduated in 2020.
Background
- Research interests are in Artificial Intelligence and Systems. Associated with SkyLab and Berkeley Artificial Intelligence Research (BAIR).
Miscellany
- Personal interests and other information not provided
Co-authors