Cargando…

Population codes enable learning from few examples by shaping inductive bias

Learning from a limited number of experiences requires suitable inductive biases. To identify how inductive biases are implemented in and shaped by neural codes, we analyze sample-efficient learning of arbitrary stimulus-response maps from arbitrary neural codes with biologically-plausible readouts....

Descripción completa

Detalles Bibliográficos
Autores principales: Bordelon, Blake, Pehlevan, Cengiz
Formato: Online Artículo Texto
Lenguaje:English
Publicado: eLife Sciences Publications, Ltd 2022
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9839349/
https://www.ncbi.nlm.nih.gov/pubmed/36524716
http://dx.doi.org/10.7554/eLife.78606
_version_ 1784869465344704512
author Bordelon, Blake
Pehlevan, Cengiz
author_facet Bordelon, Blake
Pehlevan, Cengiz
author_sort Bordelon, Blake
collection PubMed
description Learning from a limited number of experiences requires suitable inductive biases. To identify how inductive biases are implemented in and shaped by neural codes, we analyze sample-efficient learning of arbitrary stimulus-response maps from arbitrary neural codes with biologically-plausible readouts. We develop an analytical theory that predicts the generalization error of the readout as a function of the number of observed examples. Our theory illustrates in a mathematically precise way how the structure of population codes shapes inductive bias, and how a match between the code and the task is crucial for sample-efficient learning. It elucidates a bias to explain observed data with simple stimulus-response maps. Using recordings from the mouse primary visual cortex, we demonstrate the existence of an efficiency bias towards low-frequency orientation discrimination tasks for grating stimuli and low spatial frequency reconstruction tasks for natural images. We reproduce the discrimination bias in a simple model of primary visual cortex, and further show how invariances in the code to certain stimulus variations alter learning performance. We extend our methods to time-dependent neural codes and predict the sample efficiency of readouts from recurrent networks. We observe that many different codes can support the same inductive bias. By analyzing recordings from the mouse primary visual cortex, we demonstrate that biological codes have lower total activity than other codes with identical bias. Finally, we discuss implications of our theory in the context of recent developments in neuroscience and artificial intelligence. Overall, our study provides a concrete method for elucidating inductive biases of the brain and promotes sample-efficient learning as a general normative coding principle.
format Online
Article
Text
id pubmed-9839349
institution National Center for Biotechnology Information
language English
publishDate 2022
publisher eLife Sciences Publications, Ltd
record_format MEDLINE/PubMed
spelling pubmed-98393492023-01-14 Population codes enable learning from few examples by shaping inductive bias Bordelon, Blake Pehlevan, Cengiz eLife Neuroscience Learning from a limited number of experiences requires suitable inductive biases. To identify how inductive biases are implemented in and shaped by neural codes, we analyze sample-efficient learning of arbitrary stimulus-response maps from arbitrary neural codes with biologically-plausible readouts. We develop an analytical theory that predicts the generalization error of the readout as a function of the number of observed examples. Our theory illustrates in a mathematically precise way how the structure of population codes shapes inductive bias, and how a match between the code and the task is crucial for sample-efficient learning. It elucidates a bias to explain observed data with simple stimulus-response maps. Using recordings from the mouse primary visual cortex, we demonstrate the existence of an efficiency bias towards low-frequency orientation discrimination tasks for grating stimuli and low spatial frequency reconstruction tasks for natural images. We reproduce the discrimination bias in a simple model of primary visual cortex, and further show how invariances in the code to certain stimulus variations alter learning performance. We extend our methods to time-dependent neural codes and predict the sample efficiency of readouts from recurrent networks. We observe that many different codes can support the same inductive bias. By analyzing recordings from the mouse primary visual cortex, we demonstrate that biological codes have lower total activity than other codes with identical bias. Finally, we discuss implications of our theory in the context of recent developments in neuroscience and artificial intelligence. Overall, our study provides a concrete method for elucidating inductive biases of the brain and promotes sample-efficient learning as a general normative coding principle. eLife Sciences Publications, Ltd 2022-12-16 /pmc/articles/PMC9839349/ /pubmed/36524716 http://dx.doi.org/10.7554/eLife.78606 Text en © 2022, Bordelon and Pehlevan https://creativecommons.org/licenses/by/4.0/This article is distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/) , which permits unrestricted use and redistribution provided that the original author and source are credited.
spellingShingle Neuroscience
Bordelon, Blake
Pehlevan, Cengiz
Population codes enable learning from few examples by shaping inductive bias
title Population codes enable learning from few examples by shaping inductive bias
title_full Population codes enable learning from few examples by shaping inductive bias
title_fullStr Population codes enable learning from few examples by shaping inductive bias
title_full_unstemmed Population codes enable learning from few examples by shaping inductive bias
title_short Population codes enable learning from few examples by shaping inductive bias
title_sort population codes enable learning from few examples by shaping inductive bias
topic Neuroscience
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9839349/
https://www.ncbi.nlm.nih.gov/pubmed/36524716
http://dx.doi.org/10.7554/eLife.78606
work_keys_str_mv AT bordelonblake populationcodesenablelearningfromfewexamplesbyshapinginductivebias
AT pehlevancengiz populationcodesenablelearningfromfewexamplesbyshapinginductivebias