Self-Supervised Class-Cognizant Few-Shot Classification: 2022 IEEE International Conference on Image Processing (ICIP)

Ojas Kishore Shirekar; Hadi Jamali-Rad

doi:10.1109/ICIP46576.2022.9897431

Self-Supervised Class-Cognizant Few-Shot Classification: 2022 IEEE International Conference on Image Processing (ICIP)

Pattern Recognition and Bioinformatics

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

1 Citation (Scopus)

38 Downloads (Pure)

Abstract

Unsupervised learning is argued to be the dark matter of human intelligence. To build in this direction, this paper focuses on unsupervised learning from an abundance of unlabeled data followed by few-shot fine-tuning on a downstream classification task. To this aim, we extend a recent study on adopting contrastive learning for self-supervised pre-training by incorporating class-level cognizance through iterative clustering and re-ranking and by expanding the contrastive optimization loss to account for it. To our knowledge, our experimentation both in standard and cross-domain scenarios demonstrate that we set a new state-of-the-art (SoTA) in (5-way, 1 and 5-shot) settings of standard mini-ImageNet benchmark as well as the (5-way, 5 and 20-shot) settings of cross-domain CDFSL benchmark. Our code and experimentation can be found in our GitHub repository: https://github.com/ojss/c3lr.

Original language	English
Title of host publication	Proceedings of the 2022 IEEE International Conference on Image Processing (ICIP)
Place of Publication	Piscataway
Publisher	IEEE
Pages	976-980
Number of pages	5
ISBN (Electronic)	978-1-6654-9620-9
ISBN (Print)	978-1-6654-9621-6
DOIs	https://doi.org/10.1109/ICIP46576.2022.9897431
Publication status	Published - 2022
Event	2022 IEEE International Conference on Image Processing (ICIP) - Bordeaux, France Duration: 16 Oct 2022 → 19 Oct 2022

Publication series

Name	Proceedings - International Conference on Image Processing, ICIP
ISSN (Print)	1522-4880

Conference

Conference	2022 IEEE International Conference on Image Processing (ICIP)
Country/Territory	France
City	Bordeaux
Period	16/10/22 → 19/10/22

Bibliographical note

Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care
Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.

Keywords

Few-shot classification
self-supervised learning
contrastive learning

Access to Document

10.1109/ICIP46576.2022.9897431

Self-Supervised_Class-Cognizant_Few-Shot_ClassificationFinal published version, 1.82 MB

Cite this

Shirekar, O. K., & Jamali-Rad, H. (2022). Self-Supervised Class-Cognizant Few-Shot Classification: 2022 IEEE International Conference on Image Processing (ICIP). In Proceedings of the 2022 IEEE International Conference on Image Processing (ICIP) (pp. 976-980). (Proceedings - International Conference on Image Processing, ICIP). IEEE. https://doi.org/10.1109/ICIP46576.2022.9897431

@inproceedings{aeda9ccf50814cd9994d914738fad579,

title = "Self-Supervised Class-Cognizant Few-Shot Classification: 2022 IEEE International Conference on Image Processing (ICIP)",

abstract = "Unsupervised learning is argued to be the dark matter of human intelligence. To build in this direction, this paper focuses on unsupervised learning from an abundance of unlabeled data followed by few-shot fine-tuning on a downstream classification task. To this aim, we extend a recent study on adopting contrastive learning for self-supervised pre-training by incorporating class-level cognizance through iterative clustering and re-ranking and by expanding the contrastive optimization loss to account for it. To our knowledge, our experimentation both in standard and cross-domain scenarios demonstrate that we set a new state-of-the-art (SoTA) in (5-way, 1 and 5-shot) settings of standard mini-ImageNet benchmark as well as the (5-way, 5 and 20-shot) settings of cross-domain CDFSL benchmark. Our code and experimentation can be found in our GitHub repository: https://github.com/ojss/c3lr.",

keywords = "Few-shot classification, self-supervised learning, contrastive learning",

author = "Shirekar, {Ojas Kishore} and Hadi Jamali-Rad",

note = "Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public. ; 2022 IEEE International Conference on Image Processing (ICIP) ; Conference date: 16-10-2022 Through 19-10-2022",

year = "2022",

doi = "10.1109/ICIP46576.2022.9897431",

language = "English",

isbn = "978-1-6654-9621-6",

series = "Proceedings - International Conference on Image Processing, ICIP",

publisher = "IEEE",

pages = "976--980",

booktitle = "Proceedings of the 2022 IEEE International Conference on Image Processing (ICIP)",

address = "United States",

}

Shirekar, OK & Jamali-Rad, H 2022, Self-Supervised Class-Cognizant Few-Shot Classification: 2022 IEEE International Conference on Image Processing (ICIP). in Proceedings of the 2022 IEEE International Conference on Image Processing (ICIP). Proceedings - International Conference on Image Processing, ICIP, IEEE, Piscataway, pp. 976-980, 2022 IEEE International Conference on Image Processing (ICIP), Bordeaux, France, 16/10/22. https://doi.org/10.1109/ICIP46576.2022.9897431

Self-Supervised Class-Cognizant Few-Shot Classification: 2022 IEEE International Conference on Image Processing (ICIP). / Shirekar, Ojas Kishore ; Jamali-Rad, Hadi.
Proceedings of the 2022 IEEE International Conference on Image Processing (ICIP). Piscataway: IEEE, 2022. p. 976-980 (Proceedings - International Conference on Image Processing, ICIP).

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

TY - GEN

T1 - Self-Supervised Class-Cognizant Few-Shot Classification

T2 - 2022 IEEE International Conference on Image Processing (ICIP)

AU - Shirekar, Ojas Kishore

AU - Jamali-Rad, Hadi

N1 - Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.

PY - 2022

Y1 - 2022

N2 - Unsupervised learning is argued to be the dark matter of human intelligence. To build in this direction, this paper focuses on unsupervised learning from an abundance of unlabeled data followed by few-shot fine-tuning on a downstream classification task. To this aim, we extend a recent study on adopting contrastive learning for self-supervised pre-training by incorporating class-level cognizance through iterative clustering and re-ranking and by expanding the contrastive optimization loss to account for it. To our knowledge, our experimentation both in standard and cross-domain scenarios demonstrate that we set a new state-of-the-art (SoTA) in (5-way, 1 and 5-shot) settings of standard mini-ImageNet benchmark as well as the (5-way, 5 and 20-shot) settings of cross-domain CDFSL benchmark. Our code and experimentation can be found in our GitHub repository: https://github.com/ojss/c3lr.

AB - Unsupervised learning is argued to be the dark matter of human intelligence. To build in this direction, this paper focuses on unsupervised learning from an abundance of unlabeled data followed by few-shot fine-tuning on a downstream classification task. To this aim, we extend a recent study on adopting contrastive learning for self-supervised pre-training by incorporating class-level cognizance through iterative clustering and re-ranking and by expanding the contrastive optimization loss to account for it. To our knowledge, our experimentation both in standard and cross-domain scenarios demonstrate that we set a new state-of-the-art (SoTA) in (5-way, 1 and 5-shot) settings of standard mini-ImageNet benchmark as well as the (5-way, 5 and 20-shot) settings of cross-domain CDFSL benchmark. Our code and experimentation can be found in our GitHub repository: https://github.com/ojss/c3lr.

KW - Few-shot classification

KW - self-supervised learning

KW - contrastive learning

UR - http://www.scopus.com/inward/record.url?scp=85146667800&partnerID=8YFLogxK

U2 - 10.1109/ICIP46576.2022.9897431

DO - 10.1109/ICIP46576.2022.9897431

M3 - Conference contribution

SN - 978-1-6654-9621-6

T3 - Proceedings - International Conference on Image Processing, ICIP

SP - 976

EP - 980

BT - Proceedings of the 2022 IEEE International Conference on Image Processing (ICIP)

PB - IEEE

CY - Piscataway

Y2 - 16 October 2022 through 19 October 2022

ER -

Shirekar OK , Jamali-Rad H. Self-Supervised Class-Cognizant Few-Shot Classification: 2022 IEEE International Conference on Image Processing (ICIP). In Proceedings of the 2022 IEEE International Conference on Image Processing (ICIP). Piscataway: IEEE. 2022. p. 976-980. (Proceedings - International Conference on Image Processing, ICIP). doi: 10.1109/ICIP46576.2022.9897431

Self-Supervised Class-Cognizant Few-Shot Classification: 2022 IEEE International Conference on Image Processing (ICIP)

Abstract

Publication series

Conference

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this