Why state-of-the-art deep learning barely works as good as a linear classifier in extreme multi-label text classification

Mohammadreza Mohammadnia Qaraei; Sujay Khandagale; Rohit Babbar

Why state-of-the-art deep learning barely works as good as a linear classifier in extreme multi-label text classification

Mohammadreza Mohammadnia Qaraei, Sujay Khandagale, Rohit Babbar

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

Abstract

Extreme Multi-label Text Classification (XMTC) refers to supervised learning of a classifier which can predict a small subset of relevant labels for a document from an extremely large set. Even though deep learning algorithms have surpassed linear and kernel methods for most natural language processing tasks over the last decade; recent works show that state-of-the-art deep learning methods can only barely manage to work as well as a linear classifier for the XMTC task.
The goal of this work is twofold : (i) to investigate the reasons for the comparable performance of these two strands of methods for XMTC, and (ii) to document this observation explicitly, as the efficacy of linear classifiers in this regime, has been ignored in many relevant recent works.

Original language	English
Title of host publication	ESANN 2020 - Proceedings, 28th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning
Pages	223-228
Number of pages	6
ISBN (Electronic)	9782875870742
Publication status	Published - 2020
MoE publication type	A4 Article in a conference publication
Event	European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning - Virtual, Bruges, Belgium Duration: 2 Oct 2020 → 4 Oct 2020 Conference number: 28

Conference

Conference	European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning
Abbreviated title	ESANN
Country	Belgium
City	Bruges
Period	02/10/2020 → 04/10/2020
Other	Virtual

Access to Document

https://www.esann.org/sites/default/files/proceedings/2020/ES2020-207.pdf

Fingerprint Dive into the research topics of 'Why state-of-the-art deep learning barely works as good as a linear classifier in extreme multi-label text classification'. Together they form a unique fingerprint.

Cite this

Mohammadnia Qaraei, M., Khandagale, S., & Babbar, R. (2020). Why state-of-the-art deep learning barely works as good as a linear classifier in extreme multi-label text classification. In ESANN 2020 - Proceedings, 28th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning (pp. 223-228) https://www.esann.org/sites/default/files/proceedings/2020/ES2020-207.pdf

@inproceedings{f9df0ccd63f0456c97daed9ecd333cd7,

title = "Why state-of-the-art deep learning barely works as good as a linear classifier in extreme multi-label text classification",

abstract = "Extreme Multi-label Text Classification (XMTC) refers to supervised learning of a classifier which can predict a small subset of relevant labels for a document from an extremely large set. Even though deep learning algorithms have surpassed linear and kernel methods for most natural language processing tasks over the last decade; recent works show that state-of-the-art deep learning methods can only barely manage to work as well as a linear classifier for the XMTC task.The goal of this work is twofold : (i) to investigate the reasons for the comparable performance of these two strands of methods for XMTC, and (ii) to document this observation explicitly, as the efficacy of linear classifiers in this regime, has been ignored in many relevant recent works.",

author = "{Mohammadnia Qaraei}, Mohammadreza and Sujay Khandagale and Rohit Babbar",

year = "2020",

language = "English",

pages = "223--228",

booktitle = "ESANN 2020 - Proceedings, 28th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning",

note = "European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, ESANN ; Conference date: 02-10-2020 Through 04-10-2020",

}

Mohammadnia Qaraei, M, Khandagale, S & Babbar, R 2020, Why state-of-the-art deep learning barely works as good as a linear classifier in extreme multi-label text classification. in ESANN 2020 - Proceedings, 28th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning. pp. 223-228, European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, Bruges, Belgium, 02/10/2020. <https://www.esann.org/sites/default/files/proceedings/2020/ES2020-207.pdf>

Why state-of-the-art deep learning barely works as good as a linear classifier in extreme multi-label text classification. / Mohammadnia Qaraei, Mohammadreza; Khandagale, Sujay; Babbar, Rohit.

ESANN 2020 - Proceedings, 28th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning. 2020. p. 223-228.

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

TY - GEN

T1 - Why state-of-the-art deep learning barely works as good as a linear classifier in extreme multi-label text classification

AU - Mohammadnia Qaraei, Mohammadreza

AU - Khandagale, Sujay

AU - Babbar, Rohit

N1 - Conference code: 28

PY - 2020

Y1 - 2020

N2 - Extreme Multi-label Text Classification (XMTC) refers to supervised learning of a classifier which can predict a small subset of relevant labels for a document from an extremely large set. Even though deep learning algorithms have surpassed linear and kernel methods for most natural language processing tasks over the last decade; recent works show that state-of-the-art deep learning methods can only barely manage to work as well as a linear classifier for the XMTC task.The goal of this work is twofold : (i) to investigate the reasons for the comparable performance of these two strands of methods for XMTC, and (ii) to document this observation explicitly, as the efficacy of linear classifiers in this regime, has been ignored in many relevant recent works.

AB - Extreme Multi-label Text Classification (XMTC) refers to supervised learning of a classifier which can predict a small subset of relevant labels for a document from an extremely large set. Even though deep learning algorithms have surpassed linear and kernel methods for most natural language processing tasks over the last decade; recent works show that state-of-the-art deep learning methods can only barely manage to work as well as a linear classifier for the XMTC task.The goal of this work is twofold : (i) to investigate the reasons for the comparable performance of these two strands of methods for XMTC, and (ii) to document this observation explicitly, as the efficacy of linear classifiers in this regime, has been ignored in many relevant recent works.

M3 - Conference contribution

SP - 223

EP - 228

BT - ESANN 2020 - Proceedings, 28th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning

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

Y2 - 2 October 2020 through 4 October 2020

ER -