Combating False Negatives in Adversarial Imitation Learning (Student Abstract)

Autor:	Chitwan Saharia, Leonard Boussioux, Konrad Żołna, Dzmitry Bahdanau, Maxime Chevalier-Boisvert, Yoshua Bengio, David Yu-Tung Hui
Rok vydání:	2020
Předmět:	Adversarial system business.industry Computer science Sample (statistics) General Medicine Artificial intelligence Machine learning computer.software_genre business Imitation learning computer
Zdroj:	AAAI
ISSN:	2374-3468 2159-5399
DOI:	10.1609/aaai.v34i10.7272
Popis:	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 sample efficiency over the baselines by at least an order of magnitude.
Databáze:	OpenAIRE
Externí odkaz:	https://explore.openaire.eu/search/publication?articleId=doi_________::966119f3b9ac2e93512140478ca92d6f https://doi.org/10.1609/aaai.v34i10.7272 Zobrazit plný text záznamu