I'm a postdoc at UC Berkeley doing deep learning research in Pieter Abbeel's group. My primary focus areas are unsupervised learning and reinforcement learning. Before this, I founded a venture-backed startup (YC W17, F30<30) and before that got a PhD in theoretical physics from UChicago where I was a Bloomenthal Fellow.
Links: Twitter, Google Scholar, Email
Multi-head Attention, GPT, and BERT
Efficient Patch Extraction
In a series of papers on representation learning (CURL, RAD, ATC) we showed that RL from pixels can be as efficient as RL from state and even learn real-robot control policies from pixels in just 30 mins of training (FERM). These days I work on self-supervised exploration and skill extraction.
* indicates equal contribution
Hierarchical Few-Shot Imitation with Skill Transition Models Kourosh Hakhamaneshi*, Ruihan Zhao*, Albert Zhan*, Pieter Abbeel, Michael Laskin, 2021
Behavioral Priors and Dynamics Models: Improving Performance and Domain Transfer in Offline RL, Catherine Cang, Aravind Rajeswaran, Pieter Abbeel, Michael Laskin, 2021
A Framework for Efficient Robotic Manipulation Albert Zhan*, Philip Zhao*, Lerrel Pinto, Pieter Abbeel, Michael Laskin, 2021
URLB: Unsupervised Reinforcement Learning Benchmark Michael Laskin*, Denis Yarats*, Hao Liu, Kimin Lee, Albert Zhan, Kevin Lu, Catherine Cang, Lerrel Pinto, Pieter Abbeel, NeurIPS, 2021
Reinforcement Learning with Latent Flow Wenling Shang*, Xiaofei Wang*, Aravind Srinivas, Aravind Rajeswaran, Yang Gao, Pieter Abbeel, Michael Laskin, NeurIPS, 2021
Decision Transformer: Reinforcement Learning via Sequence Modeling Lili Chen, Kevin Lu, Aravind Rajeswaran, Kimin Lee, Aditya Grover, Michael Laskin, Pieter Abbeel, Aravind Srinivas, Igor Mordatch, NeurIPS, 2021