TY - JOUR
T1 - Learning efficient task-dependent representations with synaptic plasticity
AU - Bredenberg, Colin
AU - Simoncelli, Eero P.
AU - Savin, Cristina
N1 - Funding Information:
We thank the following funding sources for their support: NRT-HDR 1922658, a Google faculty award (CS), and the Howard Hughes Medical Institute (ES, CB). The authors declare no competing interests.
Publisher Copyright:
© 2020 Neural information processing systems foundation. All rights reserved.
PY - 2020
Y1 - 2020
N2 - Neural populations encode the sensory world imperfectly: their capacity is limited by the number of neurons, availability of metabolic and other biophysical resources, and intrinsic noise. The brain is presumably shaped by these limitations, improving efficiency by discarding some aspects of incoming sensory streams, while preferentially preserving commonly occurring, behaviorally-relevant information. Here we construct a stochastic recurrent neural circuit model that can learn efficient, task-specific sensory codes using a novel form of reward-modulated Hebbian synaptic plasticity. We illustrate the flexibility of the model by training an initially unstructured neural network to solve two different tasks: stimulus estimation, and stimulus discrimination. The network achieves high performance in both tasks by appropriately allocating resources and using its recurrent circuitry to best compensate for different levels of noise. We also show how the interaction between stimulus priors and task structure dictates the emergent network representations.
AB - Neural populations encode the sensory world imperfectly: their capacity is limited by the number of neurons, availability of metabolic and other biophysical resources, and intrinsic noise. The brain is presumably shaped by these limitations, improving efficiency by discarding some aspects of incoming sensory streams, while preferentially preserving commonly occurring, behaviorally-relevant information. Here we construct a stochastic recurrent neural circuit model that can learn efficient, task-specific sensory codes using a novel form of reward-modulated Hebbian synaptic plasticity. We illustrate the flexibility of the model by training an initially unstructured neural network to solve two different tasks: stimulus estimation, and stimulus discrimination. The network achieves high performance in both tasks by appropriately allocating resources and using its recurrent circuitry to best compensate for different levels of noise. We also show how the interaction between stimulus priors and task structure dictates the emergent network representations.
UR - http://www.scopus.com/inward/record.url?scp=85108421994&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85108421994&partnerID=8YFLogxK
M3 - Conference article
AN - SCOPUS:85108421994
SN - 1049-5258
VL - 2020-December
JO - Advances in Neural Information Processing Systems
JF - Advances in Neural Information Processing Systems
T2 - 34th Conference on Neural Information Processing Systems, NeurIPS 2020
Y2 - 6 December 2020 through 12 December 2020
ER -