New Reinforcement Learning Method Uses Human Cues To Correct Its Mistakes

Scientists at the University of California, Berkeley have developed a novel machine learning (ML) method, termed “reinforcement learning via intervention feedback” (RLIF), which combines reinforcement learning and interactive imitation learning to make it easier to train AI systems for complex environments. RLIF is useful in settings where a reward signal is not readily available and human feedback is not very precise, such as robotics problems. It also helps to mitigate the “distribution mismatch problem” by having experts provide real-time feedback to refine the agent’s behavior.