Exploring neural architecture search for text classification

Sheng Zhang; Lixiang Guo; Jing Fan; Xin Zhang; Weiming Zhang

doi:10.1117/12.2639851

19 October 2022 Exploring neural architecture search for text classification

Sheng Zhang, Lixiang Guo, Jing Fan, Xin Zhang, Weiming Zhang

Proceedings Volume 12294, 7th International Symposium on Advances in Electrical, Electronics, and Computer Engineering; 122945T (2022) https://doi.org/10.1117/12.2639851
Event: 7th International Symposium on Advances in Electrical, Electronics and Computer Engineering (ISAEECE 2022), 2022, Xishuangbanna, China

Abstract

Recent years have witnessed the impressive progress in applying well-designed neural networks to text classification tasks, due to their capabilities of extracting semantic information from text. However, designing such a neural network often requires a lot of expertise and repeated testing. In this paper, we aim to overcome this problem by exploring automatic neural architecture search (NAS) for text classification. We propose a genetic algorithm based one-shot neural architecture framework to automatically generate proper neural architectures for specific tasks or datasets without further re-training. The experimental results show that, for text classification, our searched neural networks achieve competitive accuracy compared with hand-crafted networks and other NAS algorithms. Our models also show promising results when transfer to some large datasets.

Citation Download Citation

Sheng Zhang, Lixiang Guo, Jing Fan, Xin Zhang, and Weiming Zhang "Exploring neural architecture search for text classification", Proc. SPIE 12294, 7th International Symposium on Advances in Electrical, Electronics, and Computer Engineering, 122945T (19 October 2022); https://doi.org/10.1117/12.2639851

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available