Ensuring AI works with the right dose of curiosity
A new algorithm makes headway in solving the problem of exploration vs. exploitation in reinforcement learning. Developed by MIT CSAIL researchers, it attempts to make decision-making more efficient.