Robust Imitation Learning from Noisy Demonstrations

Voot Tangkaratt · Nontawat Charoenphakdee · Masashi Sugiyama

Keywords: [ Reinforcement Learning ]

[ Abstract ]
Tue 13 Apr 6:30 p.m. PDT — 8:30 p.m. PDT


Robust learning from noisy demonstrations is a practical but highly challenging problem in imitation learning. In this paper, we first theoretically show that robust imitation learning can be achieved by optimizing a classification risk with a symmetric loss. Based on this theoretical finding, we then propose a new imitation learning method that optimizes the classification risk by effectively combining pseudo-labeling with co-training. Unlike existing methods, our method does not require additional labels or strict assumptions about noise distributions. Experimental results on continuous-control benchmarks show that our method is more robust compared to state-of-the-art methods.

Chat is not available.