Combating False Negatives in Adversarial Imitation Learning (Student Abstract)
AAAI, pp. 13999-14000, 2020.
We define the False Negatives problem and show that it is a significant limitation in adversarial imitation learning. We propose a method that solves the problem by leveraging the nature of goal-conditioned tasks. The method, dubbed Fake Conditioning, is tested on instruction following tasks in BabyAI environments, where it improves sampl...More
Full Text (Upload PDF)
PPT (Upload PPT)