Code

Project page and code for our ICML paper, Safe Imitation Learning via Fast Bayesian Reward Inference from Preferences. Provides code for pretraining linear reward features, performing fast Bayesian reward inference, and high-confidence policy evaluations for imitation learning.

Project page and code for our CoRL paper, Better-than-Demonstrator Imitation Learning via Automatically-Ranked Demonstrations. Provides code for learning reward functions from automatically ranked demonstrations that are generated via noise injection into a cloned policy.

Project page and code for our ICML paper, Extrapolating Beyond Suboptimal Demonstrations via Inverse Reinforcement Learning from Observations. Provides code for learning reward functions from ranked demonstrations and code to use the learned reward functions to train a RL agent on MuJoCo and Atari tasks.

Code for our AAAI 2019 paper, Machine Teaching for Inverse Reinforcement Learning: Algorithms and Applications. Provides base code for finding the minimal number of demonstrations to teach a sequential decision making task to a learner.

Code for our CoRL 2018 paper, Risk-Aware Active Inverse Reinforcement Learning. Provides code and instructions for running our GridWorld and Tabletop placement experiments.

Code for our AAAI 2018 paper, Efficient Probabilistic Performance Bounds for Inverse Reinforcement Learning. Provides code and instructions for running our GridWorld and Driving experiments.

Code for my papers on human-swarm interaction. Includes code for the base model with a flock and torus attractor, as well as code for controlling the collective behavior of a swarm by influencing only a subset of the individuals in the swarm.