arrow
返回

A data efficient transformer based on Swin Transformer

delete2023-07-30
delete7
PRE
AI
D
Dazhi Yao
Y
Yunxue Shao *
DOI:10.1007/s00371-023-02939-2delete
delete原文链接
delete原文求助
delete分享
delete收藏
摘要

摘要

En 中文
Almost all Vision Transformer-based models need to pre-train on the massive datasets and costly computation. Suppose researchers do not have enough data to train a Vision Transformer-based model or do not have powerful GPUs to implement computation for millions of labeled data. In that case, Vision Transformer-based models have no advantages over CNNs. Swin Transformer is brought forward to figure out these problems by applying the shifted window-based self-attention, which has linear computational complexity. Although Swin Transformer significantly reduces computing costs and works well on mid-size datasets, it still performs not well when it trains on a small-size dataset. In this paper, we propose a hierarchical and data-efficient Transformer based on Swin Transformer, which we call ESwin Transformer. We mainly redesigned the patch embedding module and patch merging module of Swin Transformer. We merely applied some unsophisticated convolutional components to these modules, which significantly improved performance when we trained our model on a small dataset. Our empirical results show that ESwin Transformer trained on CIFAR10/CIFAR100 with no extra data for 300 epochs achieves 97.17%/83.78% accuracy and performs better than Swin Transformer and DeiT in the same training time.
Keyword:
Computer vision
Transformer
Classification
Data efficient

期刊

Visual Computer 封面图
Visual Computer
IF:
2.9
论文数:
4.7K
被引数:
6.5K

机构

N
Nanjing Tech University
学者数:
3.7W
论文数: 2.3W
被引数: 3.9W
引用论文

引用论文

Stability of adaptive behaviors in middle-school children with autism spectrum disorders
err2007-10-01
err0
PREAI
errRobin L. Gabriels; Bonnie Jean Ivers; Dina E. Hill; John A. Agnew; John McNeill
err分享
err收藏
Biomimetic hydroxylation of aromatic compounds: Hydrogen peroxide and manganese-polyhalogenated porphyrins as a particularly good system.
err1990-01-01
err0
PREAI
errMarie-Noelle Carrier; Corinne Scheer; Pascal Gouvine; Jean-François Bartoli; Pierrette Battioni; Daniel Mansuy
err分享
err收藏
学者 查看更多内容