Christoph Dann, Yishay Mansour, Mehryar Mohri: Reinforcement Learning Can Be More Efficient with Multiple Rewards. ICML 2023: 6948-6967