Projects

Some of my recent research projects.

Deep Imitation Learning for Bimanual Manipulation

A neural network for imitation learning for robotic table lifting tasks in simulation. Our project tackled the challenge of adapting the skills learning through demonstrations to tables spawned in different locations. The paper is published in the NeurIPS 2020 Conference.

Limited Accelerated Gradient: Method and Insight (In Progress)

A novel optimization method for training gradient-based machine learning algorithms for image recognition and speech translation. This method switches between adaptive gradient method and SGD to optimize overall performance and convergence rate.