Research Interests: I am broadly interested in developing principled and practical pipelines that extend machines’ capability to learn in realistic problem settings related to perception, control, and scientific discovery. I investigate ways to represent data in various resolution and abstraction in order to control or predict the quantity of interest with efficiency and accuracy.
- 21/10 New Preprint Crystal Diffusion Variational Autoencoder for Periodic Material Generation addressing the generation of periodic material structures with physical inductive biases using a diffusion model decoder. [Accepted to NeurIPS ML4PS workshop as a contributed talk]
- 21/10 New Preprint Fragment-based Sequential Translation for Molecular Optimization that optimizes and discovers diverse molecules using an RL approach with action abstraction.[Accepted to NeurIPS AI4Science workshop]
- 21/09 New CoRL paper Learning to Jump from Pixels on using hierarchical control for agile locomotion with visual input on the MIT mini-cheetah robot. Covered at: MIT News, AZoRobotics, The Robot Report
- 21/06 New ICML paper Learning Task Informed Abstractions on enabling model-based agents to learn better policies in cluttered visual domains.