VAESim: A probabilistic approach for self-supervised prototype discovery.
Saved in:
| Title: | VAESim: A probabilistic approach for self-supervised prototype discovery. |
|---|---|
| Authors: | Ferrante, Matteo1 (AUTHOR) matteo.ferrante@uniroma2.it, Boccato, Tommaso1 (AUTHOR), Spasov, Simeon2 (AUTHOR), Duggento, Andrea1 (AUTHOR), Toschi, Nicola1,3 (AUTHOR) |
| Source: | Image & Vision Computing. Sep2023, Vol. 137, pN.PAG-N.PAG. 1p. |
| Subjects: | Supervised learning, Prototypes, Moving average process, Cluster sampling, Diagnostic imaging, Sample size (Statistics) |
| Abstract: | In medical image datasets, discrete labels are often used to describe a continuous spectrum of conditions, making unsupervised image stratification a challenging task. In this work, we propose VAESim, an architecture for image stratification based on a conditional variational autoencoder. VAESim learns a set of prototypical vectors during training, each associated with a cluster in a continuous latent space. We perform a soft assignment of each data sample to the clusters and reconstruct the sample based on a similarity measure between the sample embedding and the prototypical vectors. To update the prototypical embeddings, we use an exponential moving average of the most similar representations between actual prototypes and samples in the batch size. We test our approach on the MNIST handwritten digit dataset and the PneumoniaMNIST medical benchmark dataset, where we show that our method outperforms baselines in terms of kNN accuracy (up to + 15 % improvement in performance) and performs at par with classification models trained in a fully supervised way. Our model also outperforms current end-to-end models for unsupervised stratification. [Display omitted] • Discover prototypes and clusters in image data with VAE's new prototypes matrix. • Improve VAE's latent space for clustering, sampling, and downstream tasks. • Novel and simple approach, outperforms other deep clustering algorithms • Self-supervised training enables data generation, reconstruction, and learning • The model is benchmarked on famous images dataset, including a medical dataset. [ABSTRACT FROM AUTHOR] |
| Copyright of Image & Vision Computing is the property of Elsevier B.V. and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Engineering Source |
| FullText | Text: Availability: 0 |
|---|---|
| Header | DbId: egs DbLabel: Engineering Source An: 170044807 AccessLevel: 6 PubType: Academic Journal PubTypeId: academicJournal PreciseRelevancyScore: 0 |
| IllustrationInfo | |
| Items | – Name: Title Label: Title Group: Ti Data: VAESim: A probabilistic approach for self-supervised prototype discovery. – Name: Author Label: Authors Group: Au Data: <searchLink fieldCode="AR" term="%22Ferrante%2C+Matteo%22">Ferrante, Matteo</searchLink><relatesTo>1</relatesTo> (AUTHOR)<i> matteo.ferrante@uniroma2.it</i><br /><searchLink fieldCode="AR" term="%22Boccato%2C+Tommaso%22">Boccato, Tommaso</searchLink><relatesTo>1</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Spasov%2C+Simeon%22">Spasov, Simeon</searchLink><relatesTo>2</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Duggento%2C+Andrea%22">Duggento, Andrea</searchLink><relatesTo>1</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Toschi%2C+Nicola%22">Toschi, Nicola</searchLink><relatesTo>1,3</relatesTo> (AUTHOR) – Name: TitleSource Label: Source Group: Src Data: <searchLink fieldCode="JN" term="%22Image+%26+Vision+Computing%22">Image & Vision Computing</searchLink>. Sep2023, Vol. 137, pN.PAG-N.PAG. 1p. – Name: Subject Label: Subjects Group: Su Data: <searchLink fieldCode="DE" term="%22Supervised+learning%22">Supervised learning</searchLink><br /><searchLink fieldCode="DE" term="%22Prototypes%22">Prototypes</searchLink><br /><searchLink fieldCode="DE" term="%22Moving+average+process%22">Moving average process</searchLink><br /><searchLink fieldCode="DE" term="%22Cluster+sampling%22">Cluster sampling</searchLink><br /><searchLink fieldCode="DE" term="%22Diagnostic+imaging%22">Diagnostic imaging</searchLink><br /><searchLink fieldCode="DE" term="%22Sample+size+%28Statistics%29%22">Sample size (Statistics)</searchLink> – Name: Abstract Label: Abstract Group: Ab Data: In medical image datasets, discrete labels are often used to describe a continuous spectrum of conditions, making unsupervised image stratification a challenging task. In this work, we propose VAESim, an architecture for image stratification based on a conditional variational autoencoder. VAESim learns a set of prototypical vectors during training, each associated with a cluster in a continuous latent space. We perform a soft assignment of each data sample to the clusters and reconstruct the sample based on a similarity measure between the sample embedding and the prototypical vectors. To update the prototypical embeddings, we use an exponential moving average of the most similar representations between actual prototypes and samples in the batch size. We test our approach on the MNIST handwritten digit dataset and the PneumoniaMNIST medical benchmark dataset, where we show that our method outperforms baselines in terms of kNN accuracy (up to + 15 % improvement in performance) and performs at par with classification models trained in a fully supervised way. Our model also outperforms current end-to-end models for unsupervised stratification. [Display omitted] • Discover prototypes and clusters in image data with VAE's new prototypes matrix. • Improve VAE's latent space for clustering, sampling, and downstream tasks. • Novel and simple approach, outperforms other deep clustering algorithms • Self-supervised training enables data generation, reconstruction, and learning • The model is benchmarked on famous images dataset, including a medical dataset. [ABSTRACT FROM AUTHOR] – Name: AbstractSuppliedCopyright Label: Group: Ab Data: <i>Copyright of Image & Vision Computing is the property of Elsevier B.V. and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract.</i> (Copyright applies to all Abstracts.) |
| PLink | https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=egs&AN=170044807 |
| RecordInfo | BibRecord: BibEntity: Identifiers: – Type: doi Value: 10.1016/j.imavis.2023.104746 Languages: – Code: eng Text: English PhysicalDescription: Pagination: PageCount: 1 StartPage: N.PAG Subjects: – SubjectFull: Supervised learning Type: general – SubjectFull: Prototypes Type: general – SubjectFull: Moving average process Type: general – SubjectFull: Cluster sampling Type: general – SubjectFull: Diagnostic imaging Type: general – SubjectFull: Sample size (Statistics) Type: general Titles: – TitleFull: VAESim: A probabilistic approach for self-supervised prototype discovery. Type: main BibRelationships: HasContributorRelationships: – PersonEntity: Name: NameFull: Ferrante, Matteo – PersonEntity: Name: NameFull: Boccato, Tommaso – PersonEntity: Name: NameFull: Spasov, Simeon – PersonEntity: Name: NameFull: Duggento, Andrea – PersonEntity: Name: NameFull: Toschi, Nicola IsPartOfRelationships: – BibEntity: Dates: – D: 01 M: 09 Text: Sep2023 Type: published Y: 2023 Identifiers: – Type: issn-print Value: 02628856 Numbering: – Type: volume Value: 137 Titles: – TitleFull: Image & Vision Computing Type: main |
| ResultId | 1 |
