Data-Efficient Language Shaped Few-shot Image Classification

Zhenwen Liang, Xiangliang Zhang


Abstract
Many existing works have demonstrated that language is a helpful guider for image understanding by neural networks. We focus on a language-shaped learning problem in a few-shot setting, i.e., using language to improve few-shot image classification when language descriptions are only available during training. We propose a data-efficient method that can make the best usage of the few-shot images and the language available only in training. Experimental results on dataset ShapeWorld and Birds show that our method outperforms other state-of-the-art baselines in language-shaped few-shot learning area, especially when training data is more severely limited. Therefore, we call our approach data-efficient language-shaped learning (DF-LSL).
Anthology ID:
2021.findings-emnlp.400
Volume:
Findings of the Association for Computational Linguistics: EMNLP 2021
Month:
November
Year:
2021
Address:
Punta Cana, Dominican Republic
Venue:
Findings
SIG:
SIGDAT
Publisher:
Association for Computational Linguistics
Note:
Pages:
4680–4686
Language:
URL:
https://aclanthology.org/2021.findings-emnlp.400
DOI:
10.18653/v1/2021.findings-emnlp.400
Bibkey:
Cite (ACL):
Zhenwen Liang and Xiangliang Zhang. 2021. Data-Efficient Language Shaped Few-shot Image Classification. In Findings of the Association for Computational Linguistics: EMNLP 2021, pages 4680–4686, Punta Cana, Dominican Republic. Association for Computational Linguistics.
Cite (Informal):
Data-Efficient Language Shaped Few-shot Image Classification (Liang & Zhang, Findings 2021)
Copy Citation:
PDF:
https://aclanthology.org/2021.findings-emnlp.400.pdf
Video:
 https://aclanthology.org/2021.findings-emnlp.400.mp4
Code
 derderking/df-lsl
Data
CUB-200-2011ShapeWorld