Most of my research has been in RL or computer vision. At Uber AI, I focused on finding sample-efficient exploration methods for RL and at Berkeley, I worked on computer vision systems for autonomous driving. There, my work revolved around transfer learning, specifically from synthetic environments, and data labeling for large-scale datasets.
DECORE: Deep Compression with Reinforcement Learning
CVPR '22
Scaling Map-Elites to Deep Neuroevolution
GECCO '20
An Atari Model Zoo for Analyzing, Visualizing, and Comparing Deep Reinforcement Learning Agents
IJCAI '19
Improving Exploration in Evolution Strategies for Deep RL via a Population of Novelty-Seeking Agents
NeurIPS '18
Deep Neuroevolution: Genetic Algorithms are a Competitive Alternative for Training Deep Neural Networks for RL
Best Practices for Fine-Tuning Visual Classifiers to New Domains
ECCV '16