---
_id: '9362'
abstract:
- lang: eng
text: A central goal in systems neuroscience is to understand the functions performed
by neural circuits. Previous top-down models addressed this question by comparing
the behaviour of an ideal model circuit, optimised to perform a given function,
with neural recordings. However, this requires guessing in advance what function
is being performed, which may not be possible for many neural systems. To address
this, we propose an inverse reinforcement learning (RL) framework for inferring
the function performed by a neural network from data. We assume that the responses
of each neuron in a network are optimised so as to drive the network towards ‘rewarded’
states, that are desirable for performing a given function. We then show how one
can use inverse RL to infer the reward function optimised by the network from
observing its responses. This inferred reward function can be used to predict
how the neural network should adapt its dynamics to perform the same function
when the external environment or network structure changes. This could lead to
theoretical predictions about how neural network dynamics adapt to deal with cell
death and/or varying sensory stimulus statistics.
acknowledgement: The authors would like to thank Ulisse Ferrari for useful discussions
and feedback.
article_number: e0248940
article_processing_charge: No
article_type: original
author:
- first_name: Matthew J
full_name: Chalk, Matthew J
id: 2BAAC544-F248-11E8-B48F-1D18A9856A87
last_name: Chalk
orcid: 0000-0001-7782-4436
- first_name: Gašper
full_name: Tkačik, Gašper
id: 3D494DCA-F248-11E8-B48F-1D18A9856A87
last_name: Tkačik
orcid: 0000-0002-6699-1455
- first_name: Olivier
full_name: Marre, Olivier
last_name: Marre
citation:
ama: Chalk MJ, Tkačik G, Marre O. Inferring the function performed by a recurrent
neural network. PLoS ONE. 2021;16(4). doi:10.1371/journal.pone.0248940
apa: Chalk, M. J., Tkačik, G., & Marre, O. (2021). Inferring the function performed
by a recurrent neural network. PLoS ONE. Public Library of Science. https://doi.org/10.1371/journal.pone.0248940
chicago: Chalk, Matthew J, Gašper Tkačik, and Olivier Marre. “Inferring the Function
Performed by a Recurrent Neural Network.” PLoS ONE. Public Library of Science,
2021. https://doi.org/10.1371/journal.pone.0248940.
ieee: M. J. Chalk, G. Tkačik, and O. Marre, “Inferring the function performed by
a recurrent neural network,” PLoS ONE, vol. 16, no. 4. Public Library of
Science, 2021.
ista: Chalk MJ, Tkačik G, Marre O. 2021. Inferring the function performed by a recurrent
neural network. PLoS ONE. 16(4), e0248940.
mla: Chalk, Matthew J., et al. “Inferring the Function Performed by a Recurrent
Neural Network.” PLoS ONE, vol. 16, no. 4, e0248940, Public Library of
Science, 2021, doi:10.1371/journal.pone.0248940.
short: M.J. Chalk, G. Tkačik, O. Marre, PLoS ONE 16 (2021).
date_created: 2021-05-02T22:01:28Z
date_published: 2021-04-15T00:00:00Z
date_updated: 2023-10-18T08:17:42Z
day: '15'
ddc:
- '570'
department:
- _id: GaTk
doi: 10.1371/journal.pone.0248940
external_id:
isi:
- '000641474900072'
pmid:
- '33857170'
file:
- access_level: open_access
checksum: c52da133850307d2031f552d998f00e8
content_type: application/pdf
creator: kschuh
date_created: 2021-05-04T13:22:19Z
date_updated: 2021-05-04T13:22:19Z
file_id: '9371'
file_name: 2021_pone_Chalk.pdf
file_size: 2768282
relation: main_file
success: 1
file_date_updated: 2021-05-04T13:22:19Z
has_accepted_license: '1'
intvolume: ' 16'
isi: 1
issue: '4'
language:
- iso: eng
month: '04'
oa: 1
oa_version: Published Version
pmid: 1
publication: PLoS ONE
publication_identifier:
eissn:
- '19326203'
publication_status: published
publisher: Public Library of Science
quality_controlled: '1'
scopus_import: '1'
status: public
title: Inferring the function performed by a recurrent neural network
tmp:
image: /images/cc_by.png
legal_code_url: https://creativecommons.org/licenses/by/4.0/legalcode
name: Creative Commons Attribution 4.0 International Public License (CC-BY 4.0)
short: CC BY (4.0)
type: journal_article
user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87
volume: 16
year: '2021'
...
---
_id: '543'
abstract:
- lang: eng
text: A central goal in theoretical neuroscience is to predict the response properties
of sensory neurons from first principles. To this end, “efficient coding” posits
that sensory neurons encode maximal information about their inputs given internal
constraints. There exist, however, many variants of efficient coding (e.g., redundancy
reduction, different formulations of predictive coding, robust coding, sparse
coding, etc.), differing in their regimes of applicability, in the relevance of
signals to be encoded, and in the choice of constraints. It is unclear how these
types of efficient coding relate or what is expected when different coding objectives
are combined. Here we present a unified framework that encompasses previously
proposed efficient coding models and extends to unique regimes. We show that optimizing
neural responses to encode predictive information can lead them to either correlate
or decorrelate their inputs, depending on the stimulus statistics; in contrast,
at low noise, efficiently encoding the past always predicts decorrelation. Later,
we investigate coding of naturalistic movies and show that qualitatively different
types of visual motion tuning and levels of response sparsity are predicted, depending
on whether the objective is to recover the past or predict the future. Our approach
promises a way to explain the observed diversity of sensory neural responses,
as due to multiple functional goals and constraints fulfilled by different cell
types and/or circuits.
article_processing_charge: No
author:
- first_name: Matthew J
full_name: Chalk, Matthew J
id: 2BAAC544-F248-11E8-B48F-1D18A9856A87
last_name: Chalk
orcid: 0000-0001-7782-4436
- first_name: Olivier
full_name: Marre, Olivier
last_name: Marre
- first_name: Gasper
full_name: Tkacik, Gasper
id: 3D494DCA-F248-11E8-B48F-1D18A9856A87
last_name: Tkacik
orcid: 0000-0002-6699-1455
citation:
ama: Chalk MJ, Marre O, Tkačik G. Toward a unified theory of efficient, predictive,
and sparse coding. PNAS. 2018;115(1):186-191. doi:10.1073/pnas.1711114115
apa: Chalk, M. J., Marre, O., & Tkačik, G. (2018). Toward a unified theory of
efficient, predictive, and sparse coding. PNAS. National Academy of Sciences.
https://doi.org/10.1073/pnas.1711114115
chicago: Chalk, Matthew J, Olivier Marre, and Gašper Tkačik. “Toward a Unified Theory
of Efficient, Predictive, and Sparse Coding.” PNAS. National Academy of
Sciences, 2018. https://doi.org/10.1073/pnas.1711114115.
ieee: M. J. Chalk, O. Marre, and G. Tkačik, “Toward a unified theory of efficient,
predictive, and sparse coding,” PNAS, vol. 115, no. 1. National Academy
of Sciences, pp. 186–191, 2018.
ista: Chalk MJ, Marre O, Tkačik G. 2018. Toward a unified theory of efficient, predictive,
and sparse coding. PNAS. 115(1), 186–191.
mla: Chalk, Matthew J., et al. “Toward a Unified Theory of Efficient, Predictive,
and Sparse Coding.” PNAS, vol. 115, no. 1, National Academy of Sciences,
2018, pp. 186–91, doi:10.1073/pnas.1711114115.
short: M.J. Chalk, O. Marre, G. Tkačik, PNAS 115 (2018) 186–191.
date_created: 2018-12-11T11:47:04Z
date_published: 2018-01-02T00:00:00Z
date_updated: 2023-09-19T10:16:35Z
day: '02'
department:
- _id: GaTk
doi: 10.1073/pnas.1711114115
external_id:
isi:
- '000419128700049'
intvolume: ' 115'
isi: 1
issue: '1'
language:
- iso: eng
main_file_link:
- open_access: '1'
url: 'https://doi.org/10.1101/152660 '
month: '01'
oa: 1
oa_version: Submitted Version
page: 186 - 191
project:
- _id: 254D1A94-B435-11E9-9278-68D0E5697425
call_identifier: FWF
grant_number: P 25651-N26
name: Sensitivity to higher-order statistics in natural scenes
publication: PNAS
publication_status: published
publisher: National Academy of Sciences
publist_id: '7273'
quality_controlled: '1'
scopus_import: '1'
status: public
title: Toward a unified theory of efficient, predictive, and sparse coding
type: journal_article
user_id: c635000d-4b10-11ee-a964-aac5a93f6ac1
volume: 115
year: '2018'
...
---
_id: '680'
abstract:
- lang: eng
text: In order to respond reliably to specific features of their environment, sensory
neurons need to integrate multiple incoming noisy signals. Crucially, they also
need to compete for the interpretation of those signals with other neurons representing
similar features. The form that this competition should take depends critically
on the noise corrupting these signals. In this study we show that for the type
of noise commonly observed in sensory systems, whose variance scales with the
mean signal, sensory neurons should selectively divide their input signals by
their predictions, suppressing ambiguous cues while amplifying others. Any change
in the stimulus context alters which inputs are suppressed, leading to a deep
dynamic reshaping of neural receptive fields going far beyond simple surround
suppression. Paradoxically, these highly variable receptive fields go alongside
and are in fact required for an invariant representation of external sensory features.
In addition to offering a normative account of context-dependent changes in sensory
responses, perceptual inference in the presence of signal-dependent noise accounts
for ubiquitous features of sensory neurons such as divisive normalization, gain
control and contrast dependent temporal dynamics.
article_number: e1005582
author:
- first_name: Matthew J
full_name: Chalk, Matthew J
id: 2BAAC544-F248-11E8-B48F-1D18A9856A87
last_name: Chalk
orcid: 0000-0001-7782-4436
- first_name: Paul
full_name: Masset, Paul
last_name: Masset
- first_name: Boris
full_name: Gutkin, Boris
last_name: Gutkin
- first_name: Sophie
full_name: Denève, Sophie
last_name: Denève
citation:
ama: Chalk MJ, Masset P, Gutkin B, Denève S. Sensory noise predicts divisive reshaping
of receptive fields. PLoS Computational Biology. 2017;13(6). doi:10.1371/journal.pcbi.1005582
apa: Chalk, M. J., Masset, P., Gutkin, B., & Denève, S. (2017). Sensory noise
predicts divisive reshaping of receptive fields. PLoS Computational Biology.
Public Library of Science. https://doi.org/10.1371/journal.pcbi.1005582
chicago: Chalk, Matthew J, Paul Masset, Boris Gutkin, and Sophie Denève. “Sensory
Noise Predicts Divisive Reshaping of Receptive Fields.” PLoS Computational
Biology. Public Library of Science, 2017. https://doi.org/10.1371/journal.pcbi.1005582.
ieee: M. J. Chalk, P. Masset, B. Gutkin, and S. Denève, “Sensory noise predicts
divisive reshaping of receptive fields,” PLoS Computational Biology, vol.
13, no. 6. Public Library of Science, 2017.
ista: Chalk MJ, Masset P, Gutkin B, Denève S. 2017. Sensory noise predicts divisive
reshaping of receptive fields. PLoS Computational Biology. 13(6), e1005582.
mla: Chalk, Matthew J., et al. “Sensory Noise Predicts Divisive Reshaping of Receptive
Fields.” PLoS Computational Biology, vol. 13, no. 6, e1005582, Public Library
of Science, 2017, doi:10.1371/journal.pcbi.1005582.
short: M.J. Chalk, P. Masset, B. Gutkin, S. Denève, PLoS Computational Biology 13
(2017).
date_created: 2018-12-11T11:47:53Z
date_published: 2017-06-01T00:00:00Z
date_updated: 2023-02-23T14:10:54Z
day: '01'
ddc:
- '571'
department:
- _id: GaTk
doi: 10.1371/journal.pcbi.1005582
file:
- access_level: open_access
checksum: 796a1026076af6f4405a47d985bc7b68
content_type: application/pdf
creator: system
date_created: 2018-12-12T10:07:47Z
date_updated: 2020-07-14T12:47:40Z
file_id: '4645'
file_name: IST-2017-898-v1+1_journal.pcbi.1005582.pdf
file_size: 14555676
relation: main_file
file_date_updated: 2020-07-14T12:47:40Z
has_accepted_license: '1'
intvolume: ' 13'
issue: '6'
language:
- iso: eng
month: '06'
oa: 1
oa_version: Published Version
publication: PLoS Computational Biology
publication_identifier:
issn:
- 1553734X
publication_status: published
publisher: Public Library of Science
publist_id: '7035'
pubrep_id: '898'
quality_controlled: '1'
related_material:
record:
- id: '9855'
relation: research_data
status: public
scopus_import: 1
status: public
title: Sensory noise predicts divisive reshaping of receptive fields
tmp:
image: /images/cc_by.png
legal_code_url: https://creativecommons.org/licenses/by/4.0/legalcode
name: Creative Commons Attribution 4.0 International Public License (CC-BY 4.0)
short: CC BY (4.0)
type: journal_article
user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87
volume: 13
year: '2017'
...
---
_id: '9855'
abstract:
- lang: eng
text: Includes derivation of optimal estimation algorithm, generalisation to non-poisson
noise statistics, correlated input noise, and implementation of in a multi-layer
neural network.
article_processing_charge: No
author:
- first_name: Matthew J
full_name: Chalk, Matthew J
id: 2BAAC544-F248-11E8-B48F-1D18A9856A87
last_name: Chalk
orcid: 0000-0001-7782-4436
- first_name: Paul
full_name: Masset, Paul
last_name: Masset
- first_name: Boris
full_name: Gutkin, Boris
last_name: Gutkin
- first_name: Sophie
full_name: Denève, Sophie
last_name: Denève
citation:
ama: Chalk MJ, Masset P, Gutkin B, Denève S. Supplementary appendix. 2017. doi:10.1371/journal.pcbi.1005582.s001
apa: Chalk, M. J., Masset, P., Gutkin, B., & Denève, S. (2017). Supplementary
appendix. Public Library of Science. https://doi.org/10.1371/journal.pcbi.1005582.s001
chicago: Chalk, Matthew J, Paul Masset, Boris Gutkin, and Sophie Denève. “Supplementary
Appendix.” Public Library of Science, 2017. https://doi.org/10.1371/journal.pcbi.1005582.s001.
ieee: M. J. Chalk, P. Masset, B. Gutkin, and S. Denève, “Supplementary appendix.”
Public Library of Science, 2017.
ista: Chalk MJ, Masset P, Gutkin B, Denève S. 2017. Supplementary appendix, Public
Library of Science, 10.1371/journal.pcbi.1005582.s001.
mla: Chalk, Matthew J., et al. Supplementary Appendix. Public Library of
Science, 2017, doi:10.1371/journal.pcbi.1005582.s001.
short: M.J. Chalk, P. Masset, B. Gutkin, S. Denève, (2017).
date_created: 2021-08-10T07:05:10Z
date_published: 2017-06-01T00:00:00Z
date_updated: 2023-02-23T12:52:17Z
day: '01'
department:
- _id: GaTk
doi: 10.1371/journal.pcbi.1005582.s001
month: '06'
oa_version: Published Version
publisher: Public Library of Science
related_material:
record:
- id: '680'
relation: used_in_publication
status: public
status: public
title: Supplementary appendix
type: research_data_reference
user_id: 6785fbc1-c503-11eb-8a32-93094b40e1cf
year: '2017'
...
---
_id: '1082'
abstract:
- lang: eng
text: In many applications, it is desirable to extract only the relevant aspects
of data. A principled way to do this is the information bottleneck (IB) method,
where one seeks a code that maximises information about a relevance variable,
Y, while constraining the information encoded about the original data, X. Unfortunately
however, the IB method is computationally demanding when data are high-dimensional
and/or non-gaussian. Here we propose an approximate variational scheme for maximising
a lower bound on the IB objective, analogous to variational EM. Using this method,
we derive an IB algorithm to recover features that are both relevant and sparse.
Finally, we demonstrate how kernelised versions of the algorithm can be used to
address a broad range of problems with non-linear relation between X and Y.
alternative_title:
- Advances in Neural Information Processing Systems
author:
- first_name: Matthew J
full_name: Chalk, Matthew J
id: 2BAAC544-F248-11E8-B48F-1D18A9856A87
last_name: Chalk
orcid: 0000-0001-7782-4436
- first_name: Olivier
full_name: Marre, Olivier
last_name: Marre
- first_name: Gasper
full_name: Tkacik, Gasper
id: 3D494DCA-F248-11E8-B48F-1D18A9856A87
last_name: Tkacik
orcid: 0000-0002-6699-1455
citation:
ama: 'Chalk MJ, Marre O, Tkačik G. Relevant sparse codes with variational information
bottleneck. In: Vol 29. Neural Information Processing Systems; 2016:1965-1973.'
apa: 'Chalk, M. J., Marre, O., & Tkačik, G. (2016). Relevant sparse codes with
variational information bottleneck (Vol. 29, pp. 1965–1973). Presented at the
NIPS: Neural Information Processing Systems, Barcelona, Spain: Neural Information
Processing Systems.'
chicago: Chalk, Matthew J, Olivier Marre, and Gašper Tkačik. “Relevant Sparse Codes
with Variational Information Bottleneck,” 29:1965–73. Neural Information Processing
Systems, 2016.
ieee: 'M. J. Chalk, O. Marre, and G. Tkačik, “Relevant sparse codes with variational
information bottleneck,” presented at the NIPS: Neural Information Processing
Systems, Barcelona, Spain, 2016, vol. 29, pp. 1965–1973.'
ista: 'Chalk MJ, Marre O, Tkačik G. 2016. Relevant sparse codes with variational
information bottleneck. NIPS: Neural Information Processing Systems, Advances
in Neural Information Processing Systems, vol. 29, 1965–1973.'
mla: Chalk, Matthew J., et al. Relevant Sparse Codes with Variational Information
Bottleneck. Vol. 29, Neural Information Processing Systems, 2016, pp. 1965–73.
short: M.J. Chalk, O. Marre, G. Tkačik, in:, Neural Information Processing Systems,
2016, pp. 1965–1973.
conference:
end_date: 2016-12-10
location: Barcelona, Spain
name: 'NIPS: Neural Information Processing Systems'
start_date: 2016-12-05
date_created: 2018-12-11T11:50:03Z
date_published: 2016-12-01T00:00:00Z
date_updated: 2021-01-12T06:48:09Z
day: '01'
department:
- _id: GaTk
intvolume: ' 29'
language:
- iso: eng
main_file_link:
- open_access: '1'
url: https://arxiv.org/abs/1605.07332
month: '12'
oa: 1
oa_version: Preprint
page: 1965-1973
publication_status: published
publisher: Neural Information Processing Systems
publist_id: '6298'
quality_controlled: '1'
related_material:
link:
- relation: other
url: https://papers.nips.cc/paper/6101-relevant-sparse-codes-with-variational-information-bottleneck
scopus_import: 1
status: public
title: Relevant sparse codes with variational information bottleneck
type: conference
user_id: 3E5EF7F0-F248-11E8-B48F-1D18A9856A87
volume: 29
year: '2016'
...
---
_id: '1266'
abstract:
- lang: eng
text: 'Cortical networks exhibit ‘global oscillations’, in which neural spike times
are entrained to an underlying oscillatory rhythm, but where individual neurons
fire irregularly, on only a fraction of cycles. While the network dynamics underlying
global oscillations have been well characterised, their function is debated. Here,
we show that such global oscillations are a direct consequence of optimal efficient
coding in spiking networks with synaptic delays and noise. To avoid firing unnecessary
spikes, neurons need to share information about the network state. Ideally, membrane
potentials should be strongly correlated and reflect a ‘prediction error’ while
the spikes themselves are uncorrelated and occur rarely. We show that the most
efficient representation is when: (i) spike times are entrained to a global Gamma
rhythm (implying a consistent representation of the error); but (ii) few neurons
fire on each cycle (implying high efficiency), while (iii) excitation and inhibition
are tightly balanced. This suggests that cortical networks exhibiting such dynamics
are tuned to achieve a maximally efficient population code.'
acknowledgement: Boris Gutkin acknowledges funding by the Russian Academic Excellence
Project '5-100’.
article_number: e13824
author:
- first_name: Matthew J
full_name: Chalk, Matthew J
id: 2BAAC544-F248-11E8-B48F-1D18A9856A87
last_name: Chalk
orcid: 0000-0001-7782-4436
- first_name: Boris
full_name: Gutkin, Boris
last_name: Gutkin
- first_name: Sophie
full_name: Denève, Sophie
last_name: Denève
citation:
ama: Chalk MJ, Gutkin B, Denève S. Neural oscillations as a signature of efficient
coding in the presence of synaptic delays. eLife. 2016;5(2016JULY). doi:10.7554/eLife.13824
apa: Chalk, M. J., Gutkin, B., & Denève, S. (2016). Neural oscillations as a
signature of efficient coding in the presence of synaptic delays. ELife.
eLife Sciences Publications. https://doi.org/10.7554/eLife.13824
chicago: Chalk, Matthew J, Boris Gutkin, and Sophie Denève. “Neural Oscillations
as a Signature of Efficient Coding in the Presence of Synaptic Delays.” ELife.
eLife Sciences Publications, 2016. https://doi.org/10.7554/eLife.13824.
ieee: M. J. Chalk, B. Gutkin, and S. Denève, “Neural oscillations as a signature
of efficient coding in the presence of synaptic delays,” eLife, vol. 5,
no. 2016JULY. eLife Sciences Publications, 2016.
ista: Chalk MJ, Gutkin B, Denève S. 2016. Neural oscillations as a signature of
efficient coding in the presence of synaptic delays. eLife. 5(2016JULY), e13824.
mla: Chalk, Matthew J., et al. “Neural Oscillations as a Signature of Efficient
Coding in the Presence of Synaptic Delays.” ELife, vol. 5, no. 2016JULY,
e13824, eLife Sciences Publications, 2016, doi:10.7554/eLife.13824.
short: M.J. Chalk, B. Gutkin, S. Denève, ELife 5 (2016).
date_created: 2018-12-11T11:51:02Z
date_published: 2016-07-01T00:00:00Z
date_updated: 2021-01-12T06:49:30Z
day: '01'
ddc:
- '571'
department:
- _id: GaTk
doi: 10.7554/eLife.13824
file:
- access_level: open_access
checksum: dc52d967dc76174477bb258d84be2899
content_type: application/pdf
creator: system
date_created: 2018-12-12T10:11:20Z
date_updated: 2020-07-14T12:44:42Z
file_id: '4874'
file_name: IST-2016-700-v1+1_e13824-download.pdf
file_size: 2819055
relation: main_file
file_date_updated: 2020-07-14T12:44:42Z
has_accepted_license: '1'
intvolume: ' 5'
issue: 2016JULY
language:
- iso: eng
month: '07'
oa: 1
oa_version: Published Version
publication: eLife
publication_status: published
publisher: eLife Sciences Publications
publist_id: '6056'
pubrep_id: '700'
quality_controlled: '1'
scopus_import: 1
status: public
title: Neural oscillations as a signature of efficient coding in the presence of synaptic
delays
tmp:
image: /images/cc_by.png
legal_code_url: https://creativecommons.org/licenses/by/4.0/legalcode
name: Creative Commons Attribution 4.0 International Public License (CC-BY 4.0)
short: CC BY (4.0)
type: journal_article
user_id: 3E5EF7F0-F248-11E8-B48F-1D18A9856A87
volume: 5
year: '2016'
...