Active one-shot learning with prototypical networks

Rinu Boney; Alexander Ilin

Active one-shot learning with prototypical networks

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Scientific › peer-review

29 Downloads (Pure)

Abstract

We consider the problem of active one-shot classification where a classifier needs to adapt to new tasks by requesting labels for one example per class from (potentially many) unlabeled examples. We propose a clustering approach to the problem. The features extracted with Prototypical Networks [1] are clustered using K-means and the label for one representative sample from each cluster is requested to label the whole cluster. We demonstrate good performance of this simple active adaptation strategy using image data.

Original language	English
Title of host publication	ESANN 2019 - Proceedings, 27th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning
Publisher	European Symposium on Artificial Neural Networks (ESANN)
Pages	583-588
Number of pages	6
ISBN (Electronic)	9782875870650
Publication status	Published - 1 Jan 2019
MoE publication type	A4 Article in a conference publication
Event	European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning - Bruges, Belgium Duration: 24 Apr 2019 → 26 Apr 2019 Conference number: 27

Conference

Conference	European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning
Abbreviated title	ESANN
Country/Territory	Belgium
City	Bruges
Period	24/04/2019 → 26/04/2019

Access to Document

SCI_Active one-shot learning with prototypical networksAccepted author manuscript, 1.44 MB

OpenUrl availability

Full text

Cite this

@inproceedings{06b617a28623465287c7bc98e5c8b444,

title = "Active one-shot learning with prototypical networks",

abstract = "We consider the problem of active one-shot classification where a classifier needs to adapt to new tasks by requesting labels for one example per class from (potentially many) unlabeled examples. We propose a clustering approach to the problem. The features extracted with Prototypical Networks [1] are clustered using K-means and the label for one representative sample from each cluster is requested to label the whole cluster. We demonstrate good performance of this simple active adaptation strategy using image data.",

author = "Rinu Boney and Alexander Ilin",

year = "2019",

month = jan,

day = "1",

language = "English",

pages = "583--588",

booktitle = "ESANN 2019 - Proceedings, 27th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning",

publisher = "European Symposium on Artificial Neural Networks (ESANN)",

note = "European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, ESANN ; Conference date: 24-04-2019 Through 26-04-2019",

}

Boney, R & Ilin, A 2019, Active one-shot learning with prototypical networks. in ESANN 2019 - Proceedings, 27th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning. European Symposium on Artificial Neural Networks (ESANN), pp. 583-588, European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, Bruges, Belgium, 24/04/2019.

Active one-shot learning with prototypical networks. / Boney, Rinu; Ilin, Alexander.
ESANN 2019 - Proceedings, 27th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning. European Symposium on Artificial Neural Networks (ESANN), 2019. p. 583-588.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Scientific › peer-review

TY - GEN

T1 - Active one-shot learning with prototypical networks

AU - Boney, Rinu

AU - Ilin, Alexander

N1 - Conference code: 27

PY - 2019/1/1

Y1 - 2019/1/1

N2 - We consider the problem of active one-shot classification where a classifier needs to adapt to new tasks by requesting labels for one example per class from (potentially many) unlabeled examples. We propose a clustering approach to the problem. The features extracted with Prototypical Networks [1] are clustered using K-means and the label for one representative sample from each cluster is requested to label the whole cluster. We demonstrate good performance of this simple active adaptation strategy using image data.

AB - We consider the problem of active one-shot classification where a classifier needs to adapt to new tasks by requesting labels for one example per class from (potentially many) unlabeled examples. We propose a clustering approach to the problem. The features extracted with Prototypical Networks [1] are clustered using K-means and the label for one representative sample from each cluster is requested to label the whole cluster. We demonstrate good performance of this simple active adaptation strategy using image data.

UR - http://www.scopus.com/inward/record.url?scp=85071339221&partnerID=8YFLogxK

UR - https://www.i6doc.com/en/book/?gcoi=28001100931280

M3 - Conference contribution

SP - 583

EP - 588

BT - ESANN 2019 - Proceedings, 27th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning

PB - European Symposium on Artificial Neural Networks (ESANN)

T2 - European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning

Y2 - 24 April 2019 through 26 April 2019

ER -

Active one-shot learning with prototypical networks

Abstract

Conference

Access to Document

OpenUrl availability

Other files and links

Fingerprint

Cite this