Toward an Integration of Deep Learning and Neuroscience.

Marblestone AH, Wayne G, Kording KP. · 2016

Read Toward an Integration of Deep Learning and Neuroscience. by Marblestone AH, Wayne G, Kording KP. (2016) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc5021692.txt.

Introduction

Adam H. Marblestone, Greg Wayne, and Konrad P. Kording argue for joining deep learning with neuroscience, in Frontiers in Computational Neuroscience in 2016. Brain science studies how nerve circuits are built in fine detail. Machine learning instead tunes simple networks by brute-force correction against a cost, a measure of error. Two shifts in machine learning, richer built-in structures and layered training rules, open a bridge between the fields, the authors propose.

They suggest the brain likewise runs on cost functions, but many of them, differing by place and by stage of growth, inside a pre-built structure fitted to the problems of behavior. On this view, the brain's special systems for attention, memory, and recurrence are tools that make learning fast and data-thrifty. The paper maps which known circuit features fit this picture and sets out tests by which neuroscience could check or refine the claims.

Catalog record

CREATORMarblestone AH, Wayne G, Kording KP.
YEAR2016
COLLECTIONPAPERS
FORMATTXT
FILE SIZE281 KB
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC5021692

Browse the full library · About the Tower of Babel