FBS Colloquia No.299Center for Information and Neural Networks

Brain-information integration to endow AI with “human-like” behavior

Despite recent advances in artificial intelligence (AI), there are non-negligible behavioral differences between AI and humans even in pattern recognition in which AI exhibits splendid performance. To solve this problem, our research group has been developing an AI system to simulate the information processing in the human brain, by integrating brain information, modeled from functional MRI data, into the system. We successfully employed this system to improve the performance of the existing AI in pattern recognition tasks that require to predict human complex behaviors. In addition, as a factor that contributes to this performance improvement, we found that the integration of brain information makes AI’s internal representation closer to brain representation, including its individual differences. In this talk, I will present a series of our research progress on brain-information integration into AI, which has a great potential to realize “human-like” AI.