Entropy, Free Energy, and Work of Restricted Boltzmann Machines

<p dir="ltr">A restricted Boltzmann machine is a generative probabilistic graphic network. A probability of finding the network in a certain configuration is given by the Boltzmann distribution. Given training data, its learning is done by optimizing the parameters of the energy func...

وصف كامل

محفوظ في:
التفاصيل البيبلوغرافية
المؤلف الرئيسي: Sangchul Oh (10980923) (author)
مؤلفون آخرون: Abdelkader Baggag (16864140) (author), Hyunchul Nha (17871053) (author)
منشور في: 2020
الموضوعات:
الوسوم: إضافة وسم
لا توجد وسوم, كن أول من يضع وسما على هذه التسجيلة!
_version_ 1864513511346929664
author Sangchul Oh (10980923)
author2 Abdelkader Baggag (16864140)
Hyunchul Nha (17871053)
author2_role author
author
author_facet Sangchul Oh (10980923)
Abdelkader Baggag (16864140)
Hyunchul Nha (17871053)
author_role author
dc.creator.none.fl_str_mv Sangchul Oh (10980923)
Abdelkader Baggag (16864140)
Hyunchul Nha (17871053)
dc.date.none.fl_str_mv 2020-05-11T09:00:00Z
dc.identifier.none.fl_str_mv 10.3390/e22050538
dc.relation.none.fl_str_mv https://figshare.com/articles/journal_contribution/Entropy_Free_Energy_and_Work_of_Restricted_Boltzmann_Machines/26176837
dc.rights.none.fl_str_mv CC BY 4.0
info:eu-repo/semantics/openAccess
dc.subject.none.fl_str_mv Information and computing sciences
Machine learning
Software engineering
Mathematical sciences
Statistics
restricted Boltzmann machines
entropy
subadditivity of entropy
Jarzynski equality
machine learning
dc.title.none.fl_str_mv Entropy, Free Energy, and Work of Restricted Boltzmann Machines
dc.type.none.fl_str_mv Text
Journal contribution
info:eu-repo/semantics/publishedVersion
text
contribution to journal
description <p dir="ltr">A restricted Boltzmann machine is a generative probabilistic graphic network. A probability of finding the network in a certain configuration is given by the Boltzmann distribution. Given training data, its learning is done by optimizing the parameters of the energy function of the network. In this paper, we analyze the training process of the restricted Boltzmann machine in the context of statistical physics. As an illustration, for small size bar-and-stripe patterns, we calculate thermodynamic quantities such as entropy, free energy, and internal energy as a function of the training epoch. We demonstrate the growth of the correlation between the visible and hidden layers via the subadditivity of entropies as the training proceeds. Using the Monte-Carlo simulation of trajectories of the visible and hidden vectors in the configuration space, we also calculate the distribution of the work done on the restricted Boltzmann machine by switching the parameters of the energy function. We discuss the Jarzynski equality which connects the path average of the exponential function of the work and the difference in free energies before and after training.</p><h2>Other Information</h2><p dir="ltr">Published in: Entropy<br>License: <a href="https://creativecommons.org/licenses/by/4.0/" target="_blank">https://creativecommons.org/licenses/by/4.0/</a><br>See article on publisher's website: <a href="https://dx.doi.org/10.3390/e22050538" target="_blank">https://dx.doi.org/10.3390/e22050538</a></p><p dir="ltr">Additional institutions affiliated with: Arts and Sciences Program - TAMUQ</p>
eu_rights_str_mv openAccess
id Manara2_b8f22daa17cef3476163cb43947cd983
identifier_str_mv 10.3390/e22050538
network_acronym_str Manara2
network_name_str Manara2
oai_identifier_str oai:figshare.com:article/26176837
publishDate 2020
repository.mail.fl_str_mv
repository.name.fl_str_mv
repository_id_str
rights_invalid_str_mv CC BY 4.0
spelling Entropy, Free Energy, and Work of Restricted Boltzmann MachinesSangchul Oh (10980923)Abdelkader Baggag (16864140)Hyunchul Nha (17871053)Information and computing sciencesMachine learningSoftware engineeringMathematical sciencesStatisticsrestricted Boltzmann machinesentropysubadditivity of entropyJarzynski equalitymachine learning<p dir="ltr">A restricted Boltzmann machine is a generative probabilistic graphic network. A probability of finding the network in a certain configuration is given by the Boltzmann distribution. Given training data, its learning is done by optimizing the parameters of the energy function of the network. In this paper, we analyze the training process of the restricted Boltzmann machine in the context of statistical physics. As an illustration, for small size bar-and-stripe patterns, we calculate thermodynamic quantities such as entropy, free energy, and internal energy as a function of the training epoch. We demonstrate the growth of the correlation between the visible and hidden layers via the subadditivity of entropies as the training proceeds. Using the Monte-Carlo simulation of trajectories of the visible and hidden vectors in the configuration space, we also calculate the distribution of the work done on the restricted Boltzmann machine by switching the parameters of the energy function. We discuss the Jarzynski equality which connects the path average of the exponential function of the work and the difference in free energies before and after training.</p><h2>Other Information</h2><p dir="ltr">Published in: Entropy<br>License: <a href="https://creativecommons.org/licenses/by/4.0/" target="_blank">https://creativecommons.org/licenses/by/4.0/</a><br>See article on publisher's website: <a href="https://dx.doi.org/10.3390/e22050538" target="_blank">https://dx.doi.org/10.3390/e22050538</a></p><p dir="ltr">Additional institutions affiliated with: Arts and Sciences Program - TAMUQ</p>2020-05-11T09:00:00ZTextJournal contributioninfo:eu-repo/semantics/publishedVersiontextcontribution to journal10.3390/e22050538https://figshare.com/articles/journal_contribution/Entropy_Free_Energy_and_Work_of_Restricted_Boltzmann_Machines/26176837CC BY 4.0info:eu-repo/semantics/openAccessoai:figshare.com:article/261768372020-05-11T09:00:00Z
spellingShingle Entropy, Free Energy, and Work of Restricted Boltzmann Machines
Sangchul Oh (10980923)
Information and computing sciences
Machine learning
Software engineering
Mathematical sciences
Statistics
restricted Boltzmann machines
entropy
subadditivity of entropy
Jarzynski equality
machine learning
status_str publishedVersion
title Entropy, Free Energy, and Work of Restricted Boltzmann Machines
title_full Entropy, Free Energy, and Work of Restricted Boltzmann Machines
title_fullStr Entropy, Free Energy, and Work of Restricted Boltzmann Machines
title_full_unstemmed Entropy, Free Energy, and Work of Restricted Boltzmann Machines
title_short Entropy, Free Energy, and Work of Restricted Boltzmann Machines
title_sort Entropy, Free Energy, and Work of Restricted Boltzmann Machines
topic Information and computing sciences
Machine learning
Software engineering
Mathematical sciences
Statistics
restricted Boltzmann machines
entropy
subadditivity of entropy
Jarzynski equality
machine learning