EMI: Exploration with Mutual Information Maximizing State and Action Embeddings

Open in new window