ENS, room Dussane, 45 rue d'Ulm, 75005 Paris
A central issue in machine and reinforcement learning is the 'curse-of-dimensionality', which arises when the dimensionality of the task/stimulus space is much larger than available training opportunities. How does the brain solve this issue? In this talk I will introduce our recent work combining human behaviour, neuroimaging and computational modelling to investigate the mechanisms that underpin learning from small samples. I will discuss and speculate how the brain may use cognitive functions to construct low dimensional task state representations on which learning can efficiently operate. This talk is based on joint work with Mitsuo Kawato, Lennart Bramlage, Benedetto De Martino and Hakwan Lau.
Aurelio Cortese is a Principal Investigator at ATR Computational Neuroscience Labs in Kyoto, Japan.