Predicting the generalization gap in neural networks using topological data analysis
Understanding how neural networks generalize on unseen data is crucial for designing more robust and reliable models. In this paper, we study the generalization gap of neural networks using methods from topological data analysis. For this purpose, we compute homological persistence diagrams of weigh...
Saved in:
Main Authors | , , , , , |
---|---|
Format | Journal Article |
Language | English |
Published |
23.03.2022
|
Subjects | |
Online Access | Get full text |
Cover
Loading…
Abstract | Understanding how neural networks generalize on unseen data is crucial for
designing more robust and reliable models. In this paper, we study the
generalization gap of neural networks using methods from topological data
analysis. For this purpose, we compute homological persistence diagrams of
weighted graphs constructed from neuron activation correlations after a
training phase, aiming to capture patterns that are linked to the
generalization capacity of the network. We compare the usefulness of different
numerical summaries from persistence diagrams and show that a combination of
some of them can accurately predict and partially explain the generalization
gap without the need of a test set. Evaluation on two computer vision
recognition tasks (CIFAR10 and SVHN) shows competitive generalization gap
prediction when compared against state-of-the-art methods. |
---|---|
AbstractList | Understanding how neural networks generalize on unseen data is crucial for
designing more robust and reliable models. In this paper, we study the
generalization gap of neural networks using methods from topological data
analysis. For this purpose, we compute homological persistence diagrams of
weighted graphs constructed from neuron activation correlations after a
training phase, aiming to capture patterns that are linked to the
generalization capacity of the network. We compare the usefulness of different
numerical summaries from persistence diagrams and show that a combination of
some of them can accurately predict and partially explain the generalization
gap without the need of a test set. Evaluation on two computer vision
recognition tasks (CIFAR10 and SVHN) shows competitive generalization gap
prediction when compared against state-of-the-art methods. |
Author | Ballester, Rubén Casacuberta, Carles Madadi, Meysam Escalera, Sergio Corneanu, Ciprian A Clemente, Xavier Arnal |
Author_xml | – sequence: 1 givenname: Rubén surname: Ballester fullname: Ballester, Rubén – sequence: 2 givenname: Xavier Arnal surname: Clemente fullname: Clemente, Xavier Arnal – sequence: 3 givenname: Carles surname: Casacuberta fullname: Casacuberta, Carles – sequence: 4 givenname: Meysam surname: Madadi fullname: Madadi, Meysam – sequence: 5 givenname: Ciprian A surname: Corneanu fullname: Corneanu, Ciprian A – sequence: 6 givenname: Sergio surname: Escalera fullname: Escalera, Sergio |
BackLink | https://doi.org/10.48550/arXiv.2203.12330$$DView paper in arXiv |
BookMark | eNotj7tOxDAURF1AAQsfQIV_IOHazjpOiVa8pJWgSB9dPxIsgh3ZWWD5ekKgOtLMaKRzTk5CDI6QKwZlpbZbuMH05T9KzkGUjAsBZ6R9Sc56M_sw0PnV0cEFl3D03zj7GOiAE_WBBndYwgXzZ0xvmR7yuo9THOPgzVJZnJFiwPGYfb4gpz2O2V3-c0Pa-7t291jsnx-edrf7AmUNhQasjYTaWJBSW4u6V32ja91YJhqFRnFwuoa-kVWjFLIKuNUMK66lNeDEhlz_3a5a3ZT8O6Zj96vXrXriB0SNTmk |
ContentType | Journal Article |
Copyright | http://arxiv.org/licenses/nonexclusive-distrib/1.0 |
Copyright_xml | – notice: http://arxiv.org/licenses/nonexclusive-distrib/1.0 |
DBID | AKY AKZ GOX |
DOI | 10.48550/arxiv.2203.12330 |
DatabaseName | arXiv Computer Science arXiv Mathematics arXiv.org |
DatabaseTitleList | |
Database_xml | – sequence: 1 dbid: GOX name: arXiv.org url: http://arxiv.org/find sourceTypes: Open Access Repository |
DeliveryMethod | fulltext_linktorsrc |
ExternalDocumentID | 2203_12330 |
GroupedDBID | AKY AKZ GOX |
ID | FETCH-LOGICAL-a670-b0a7c607cd066bddabf8f9b7b9d1398ac820eb70f964988a1402db1a42b6dc0e3 |
IEDL.DBID | GOX |
IngestDate | Mon Jan 08 05:37:20 EST 2024 |
IsDoiOpenAccess | true |
IsOpenAccess | true |
IsPeerReviewed | false |
IsScholarly | false |
Language | English |
LinkModel | DirectLink |
MergedId | FETCHMERGED-LOGICAL-a670-b0a7c607cd066bddabf8f9b7b9d1398ac820eb70f964988a1402db1a42b6dc0e3 |
OpenAccessLink | https://arxiv.org/abs/2203.12330 |
ParticipantIDs | arxiv_primary_2203_12330 |
PublicationCentury | 2000 |
PublicationDate | 2022-03-23 |
PublicationDateYYYYMMDD | 2022-03-23 |
PublicationDate_xml | – month: 03 year: 2022 text: 2022-03-23 day: 23 |
PublicationDecade | 2020 |
PublicationYear | 2022 |
Score | 1.8421015 |
SecondaryResourceType | preprint |
Snippet | Understanding how neural networks generalize on unseen data is crucial for
designing more robust and reliable models. In this paper, we study the... |
SourceID | arxiv |
SourceType | Open Access Repository |
SubjectTerms | Computer Science - Learning Mathematics - Algebraic Topology |
Title | Predicting the generalization gap in neural networks using topological data analysis |
URI | https://arxiv.org/abs/2203.12330 |
hasFullText | 1 |
inHoldings | 1 |
isFullTextHit | |
isPrint | |
link | http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwdV09T8MwFLRKJxYEAlQ-5YHV4NiuHY8IUSokPoYgZav8YqfqEqq0IH4-z3YQLKzOy3Kxcvf0zmdCrhzuEmt9wVCMKqas9gy0gCjkpshOyrcpSunpWc_f1GM9rUeE_pyFcf3X6jPnA8PmRggur_HfKrEp3xEiWrYeXuo8nExRXEP9bx1qzLT0hyRm-2RvUHf0Nn-OAzIK3SGpXvs4DYn-Yopyiy5z0vNwAJIu3ZquOhqDJfHVLtuyNzQa0rE-32EQkaTRzEndECJyRKrZfXU3Z8NlBsxpwxlwZxrNTeOR48F7B23ZWjCAOElbugaZOIDhrdXKlqXDvkd4KJwSoH3Dgzwm4-69CxNCW2N1U5RB6QCqlYUTAUklTlS9dV7xEzJJECzWOa9iEdFZJHRO_390RnZFdPZzyYQ8J-Nt_xEukG-3cJlA_wbFmIEg |
link.rule.ids | 228,230,786,891 |
linkProvider | Cornell University |
openUrl | ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Predicting+the+generalization+gap+in+neural+networks+using+topological+data+analysis&rft.au=Ballester%2C+Rub%C3%A9n&rft.au=Clemente%2C+Xavier+Arnal&rft.au=Casacuberta%2C+Carles&rft.au=Madadi%2C+Meysam&rft.date=2022-03-23&rft_id=info:doi/10.48550%2Farxiv.2203.12330&rft.externalDocID=2203_12330 |