Social diversity and social preferences in mixed-motive reinforcement learning
2020·Arxiv