Robust Imitation Learning from Noisy Demonstrations

TL;DR

This paper introduces a robust imitation learning method that effectively handles noisy demonstrations by optimizing classification risk with a symmetric loss, combining pseudo-labeling and co-training without needing extra labels or assumptions.

Contribution

The paper provides a theoretical foundation for robust imitation learning using symmetric loss and proposes a novel method combining pseudo-labeling with co-training that outperforms existing approaches.

Findings

Our method is more robust than state-of-the-art methods on continuous-control benchmarks.

It does not require additional labels or strict noise assumptions.

Theoretical analysis supports the effectiveness of symmetric loss in robust imitation learning.

Abstract

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.