Smooth Imitation Learning for Online Sequence Prediction
Hoang Le (California Institute of Technology)
Andrew Kang (California Institute of Technology)
Yisong Yue (California Institute of Technology)
Peter Carr (Disney Research Pittsburgh)
International Conference on Machine Learning (ICML) 2016
June 19, 2016
We study the problem of smooth imitation learning for online sequence prediction, where the goal is to train a policy that can smoothly imitate demonstrated behavior in a dynamic and continuous environment in response to online, sequential context input. Since the mapping from context to behavior is often complex, we take a learning reduction approach to reduce smooth imitation learning to a regression problem using complex function classes that are regularized to ensure smoothness. We present a learning meta-algorithm that achieves fast and stable convergence to a good policy. Our approach enjoys several attractive properties, including being fully deterministic, employing an adaptive learning rate that can provably yield larger policy improvements compared to previous approaches, and the ability to ensure stable convergence. Our empirical results demonstrate significant performance gains over previous approaches.
The dataset contains 47 training sequences and 2 test sequences, encoded as tab-delimited text files. Columns 1-14 represent the feature elements at each time instant, and column 15 is the broadcast camera pan angle (in degrees).
Download File "Smooth Imitation Learning for Online Sequence Prediction-Paper"
[pdf, 942.40 KB]