论文标题
从嘈杂的示范中学习的强大模仿
Robust Imitation Learning from Noisy Demonstrations
论文作者
论文摘要
从嘈杂的演示中学习的强大学习是模仿学习中的实用但高度挑战的问题。在本文中，我们首先从理论上表明可以通过以对称损失优化分类风险来实现鲁棒的模仿学习。基于这一理论发现，我们提出了一种新的模仿学习方法，该方法通过有效地将伪标记与共同培训相结合来优化分类风险。与现有方法不同，我们的方法不需要有关噪声分布的其他标签或严格的假设。连续控制基准的实验结果表明，与最新方法相比，我们的方法更强大。
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.
