Adaptive Fourier Neural Operators: Efficient Token Mixers for Transformers

计算机科学 人工智能 安全性令牌 频域 算法 理论计算机科学 计算机视觉 计算机安全
作者
John Guibas,Morteza Mardani,Zongyi Li,Andrew Tao,Anima Anandkumar,Bryan Catanzaro
出处
期刊:Cornell University - arXiv 被引量:52
标识
DOI:10.48550/arxiv.2111.13587
摘要

Vision transformers have delivered tremendous success in representation learning. This is primarily due to effective token mixing through self attention. However, this scales quadratically with the number of pixels, which becomes infeasible for high-resolution inputs. To cope with this challenge, we propose Adaptive Fourier Neural Operator (AFNO) as an efficient token mixer that learns to mix in the Fourier domain. AFNO is based on a principled foundation of operator learning which allows us to frame token mixing as a continuous global convolution without any dependence on the input resolution. This principle was previously used to design FNO, which solves global convolution efficiently in the Fourier domain and has shown promise in learning challenging PDEs. To handle challenges in visual representation learning such as discontinuities in images and high resolution inputs, we propose principled architectural modifications to FNO which results in memory and computational efficiency. This includes imposing a block-diagonal structure on the channel mixing weights, adaptively sharing weights across tokens, and sparsifying the frequency modes via soft-thresholding and shrinkage. The resulting model is highly parallel with a quasi-linear complexity and has linear memory in the sequence size. AFNO outperforms self-attention mechanisms for few-shot segmentation in terms of both efficiency and accuracy. For Cityscapes segmentation with the Segformer-B3 backbone, AFNO can handle a sequence size of 65k and outperforms other efficient self-attention mechanisms.
最长约 10秒,即可获得该文献文件

科研通智能强力驱动
Strongly Powered by AbleSci AI
更新
PDF的下载单位、IP信息已删除 (2025-6-4)

科研通是完全免费的文献互助平台,具备全网最快的应助速度,最高的求助完成率。 对每一个文献求助,科研通都将尽心尽力,给求助人一个满意的交代。
实时播报
1秒前
Evooolet完成签到,获得积分10
1秒前
ding应助凡是过往皆为序章采纳,获得10
3秒前
干净海秋完成签到,获得积分10
3秒前
小蘑菇应助zd采纳,获得10
3秒前
Ava应助别偷我增肌粉采纳,获得30
3秒前
abc完成签到,获得积分10
4秒前
一点通发布了新的文献求助10
5秒前
8秒前
淡淡的豁应助棒棒采纳,获得150
10秒前
小秋完成签到,获得积分10
10秒前
华仔应助木木林采纳,获得10
10秒前
QIZH发布了新的文献求助10
10秒前
FashionBoy应助py999采纳,获得10
11秒前
小马甲应助圆蓬蓬采纳,获得10
12秒前
12秒前
Orange应助陈念采纳,获得10
13秒前
13秒前
热心又蓝完成签到,获得积分10
14秒前
111发布了新的文献求助10
14秒前
曾经晓亦发布了新的文献求助20
14秒前
量子星尘发布了新的文献求助10
14秒前
FashionBoy应助一点通采纳,获得10
14秒前
害怕的冬云完成签到,获得积分10
15秒前
丘比特应助nni采纳,获得20
17秒前
陈一完成签到 ,获得积分10
17秒前
现代完成签到,获得积分10
18秒前
尹天奇发布了新的文献求助10
18秒前
18秒前
SYLH应助QIZH采纳,获得10
19秒前
结实的泥猴桃完成签到 ,获得积分10
20秒前
tex关闭了tex文献求助
21秒前
bkagyin应助薯条派采纳,获得10
21秒前
21秒前
21秒前
long发布了新的文献求助10
21秒前
麓麓菌完成签到 ,获得积分10
22秒前
111完成签到,获得积分10
22秒前
23秒前
向上完成签到,获得积分10
23秒前
高分求助中
A new approach to the extrapolation of accelerated life test data 1000
Picture Books with Same-sex Parented Families: Unintentional Censorship 700
ACSM’s Guidelines for Exercise Testing and Prescription, 12th edition 500
Nucleophilic substitution in azasydnone-modified dinitroanisoles 500
不知道标题是什么 500
Indomethacinのヒトにおける経皮吸収 400
Phylogenetic study of the order Polydesmida (Myriapoda: Diplopoda) 370
热门求助领域 (近24小时)
化学 材料科学 医学 生物 工程类 有机化学 生物化学 物理 内科学 纳米技术 计算机科学 化学工程 复合材料 遗传学 基因 物理化学 催化作用 冶金 细胞生物学 免疫学
热门帖子
关注 科研通微信公众号,转发送积分 3975755
求助须知:如何正确求助?哪些是违规求助? 3520108
关于积分的说明 11200829
捐赠科研通 3256492
什么是DOI,文献DOI怎么找? 1798298
邀请新用户注册赠送积分活动 877509
科研通“疑难数据库(出版商)”最低求助积分说明 806403