Stochastic Thermodynamics of Learning.

@article{Goldt2017StochasticTO,
  title={Stochastic Thermodynamics of Learning.},
  author={Sebastian Goldt and U. Seifert},
  journal={Physical review letters},
  year={2017},
  volume={118 1},
  pages={010601}
}
Virtually every organism gathers information about its noisy environment and builds models from those data, mostly using neural networks. Here, we use stochastic thermodynamics to analyze the learning of a classification rule by a neural network. We show that the information acquired by the network is bounded by the thermodynamic cost of learning and introduce a learning efficiency η≤1. We discuss the conditions for optimal learning and analyze Hebbian learning in the thermodynamic limit. 
Recent Discussions
This paper has been referenced on Twitter 36 times over the past 90 days. VIEW TWEETS

From This Paper

Figures, tables, and topics from this paper.
1 Citations
22 References
Similar Papers

Citations

Publications citing this paper.

References

Publications referenced by this paper.
Showing 1-10 of 22 references

Rep

  • U. Seifert
  • Prog. Phys. 75, 126001
  • 2012
Highly Influential
4 Excerpts

Statistical Mechanics of Learning (Cambridge

  • A. Engel, C. Van den Broeck
  • 2001
Highly Influential
5 Excerpts

and Others

  • E. R. Kandel, J. H. Schwartz, T. M. Jessell
  • Principles of Neural Science
  • 2000
Highly Influential
4 Excerpts

Stochastic Processes in Physics and Chemistry (Elsevier

  • N. van Kampen
  • New York,
  • 1992
Highly Influential
4 Excerpts

Phys

  • T. R. Sokolowski, G. Tkačik
  • Rev. E 91, 062710 (2015). PRL 118, 010601
  • 2017

Nat

  • S. Ito, T. Sagawa
  • Commun. 6, 7498
  • 2015

New J

  • J. M. Horowitz, H. Sandberg
  • Phys. 16, 125007
  • 2014

PLoS Comput

  • P. Sartori, L. Granger, C. F. Lee, J. M. Horowitz
  • Biol. 10, e1003974
  • 2014

Proc

  • A. Murugan, D. A. Huse, S. Leibler
  • Natl. Acad. Sci. U.S.A. 109, 12034
  • 2012

Similar Papers

Loading similar papers…