Sparse coding in brains and machines


Bruno Olshausen


Play Video

Play Audio File

Date: February 26, 2014


The principle of sparse coding figures prominently in both theories of brain function and in the design of signal processing and computer vision systems.  Here I will review what we know about sparse coding in brains, and some of the pressing questions that could be resolved by the new recording technologies now emerging.  I will then turn to what we have learned from technological applications about the utility of sparse coding, and what this may tell us about neural representation.  In particular, the emerging paradigm of ‘deep networks’ offers an opportunity for interdisciplinary exploration into theories of neural coding, where insights gained from one field of inquiry may inform the other.

Created: Friday, February 28th, 2014