% IMPORTANT: The following is UTF-8 encoded. This means that in the presence
% of non-ASCII characters, it will not work with BibTeX 0.99 or older.
% Instead, you should use an up-to-date BibTeX implementation like “bibtex8” or
% “biber”.
@INPROCEEDINGS{Helias:890928,
author = {Helias, Moritz and van Meegen, Alexander and Dahmen, David
and Keup, Christian and Nestler, Sandra},
title = {{F}luctuations, correlations, chaos: dynamics and
computation in recurrent networks},
reportid = {FZJ-2021-01253},
year = {2021},
abstract = {The remarkable properties of information-processing by
biological and artificial neuronal networks arise from the
interaction of large numbers of neurons. A central quest is
thus to characterize their collective states. The directed
coupling between pairs of neurons and their continuous
dissipation of energy, moreover, cause dynamics of neuronal
networks outside thermodynamic equilibrium. Tools from
non-equilibrium statistical mechanics and field theory are
thus useful to obtain a quantitative understanding. We here
present recent progress using such approaches [1].We show
how activity in large, random networks can be described by a
unified approach of path-integrals and large deviation
theory that allows the inference of parameters from data and
the prediction of future activity [2]. This approach also
allows one to quantify fluctuations around the mean-field
theory. These are important to understand why correlations
observed between pairs of neurons indicate dynamics of
cortical networks that are poised near a critical point [3].
Close to this transition, we find chaotic dynamics and
prolonged sequential memory for past signals [4]. In the
chaotic regime, networks offer representations of
information whose dimensionality expands with time. We show
how this mechanism aids classification performance [5].
Performance in such settings of reservoir computing,
moreover, sensitively depends on the way information is fed
into the network. Formally unrolling recurrence with the
help of Green‘s functions yields a controlled practical
method to optimize reservoir computing [6].Together these
works illustrate the fruitful interplay between theoretical
physics, neuronal networks, and neural information
processing.References: 1. Helias, Dahmen (2020) Statistical
field theory for neural networks. Springer lecture notes in
physics.2. Meegen, Kuehn, Helias (2020) Large Deviation
Approach to Random Recurrent Neuronal Networks: Rate
Function, Parameter Inference, and Activity Prediction
arXiv:2009.088893. Dahmen, Grün, Diesmann, Helias (2019).
Second type of criticality in the brain uncovers rich
multiple-neuron dynamics. PNAS 116 (26) 13051-130604.
Schuecker J, Goedeke S, Helias M (2018). Optimal sequence
memory in driven random networks. Phys Rev X 8, 0410295.
Keup, Kuehn, Dahmen, Helias (2020) Transient chaotic
dimensionality expansion by recurrent networks.
arXiv:2002.110066. Nestler, Keup, Dahmen, Gilson, Rauhut,
Helias (2020) Unfolding recurrence by Green's functions for
optimized reservoir computing. In Advances in Neural
Information Processing Systems 33 (NeurIPS 2020)},
organization = {MILA Seminar, online (Canada)},
subtyp = {Invited},
cin = {INM-6 / INM-10 / IAS-6},
cid = {I:(DE-Juel1)INM-6-20090406 / I:(DE-Juel1)INM-10-20170113 /
I:(DE-Juel1)IAS-6-20130828},
pnm = {5231 - Neuroscientific Foundations (POF4-523) / 5232 -
Computational Principles (POF4-523) / 5234 - Emerging NC
Architectures (POF4-523) / MSNN - Theory of multi-scale
neuronal networks (HGF-SMHB-2014-2018) / HBP SGA2 - Human
Brain Project Specific Grant Agreement 2 (785907) / HBP SGA3
- Human Brain Project Specific Grant Agreement 3 (945539) /
neuroIC002 - Recurrence and stochasticity for neuro-inspired
computation (EXS-SF-neuroIC002) / RenormalizedFlows -
Transparent Deep Learning with Renormalized Flows
(BMBF-01IS19077A) / Advanced Computing Architectures
$(aca_20190115)$ / SDS005 - Towards an integrated data
science of complex natural systems (PF-JARA-SDS005)},
pid = {G:(DE-HGF)POF4-5231 / G:(DE-HGF)POF4-5232 /
G:(DE-HGF)POF4-5234 / G:(DE-Juel1)HGF-SMHB-2014-2018 /
G:(EU-Grant)785907 / G:(EU-Grant)945539 /
G:(DE-82)EXS-SF-neuroIC002 / G:(DE-Juel-1)BMBF-01IS19077A /
$G:(DE-Juel1)aca_20190115$ / G:(DE-Juel-1)PF-JARA-SDS005},
typ = {PUB:(DE-HGF)31},
url = {https://juser.fz-juelich.de/record/890928},
}