通过减少模内重叠进行 CLIP 适应

Alexey Kravets, Vinay Namboodiri
{"title":"通过减少模内重叠进行 CLIP 适应","authors":"Alexey Kravets, Vinay Namboodiri","doi":"arxiv-2409.11338","DOIUrl":null,"url":null,"abstract":"Numerous methods have been proposed to adapt a pre-trained foundational CLIP\nmodel for few-shot classification. As CLIP is trained on a large corpus, it\ngeneralises well through adaptation to few-shot classification. In this work,\nwe analyse the intra-modal overlap in image space in terms of embedding\nrepresentation. Our analysis shows that, due to contrastive learning,\nembeddings from CLIP model exhibit high cosine similarity distribution overlap\nin the image space between paired and unpaired examples affecting the\nperformance of few-shot training-free classification methods which rely on\nsimilarity in the image space for their predictions. To tackle intra-modal\noverlap we propose to train a lightweight adapter on a generic set of samples\nfrom the Google Open Images dataset demonstrating that this improves accuracy\nfor few-shot training-free classification. We validate our contribution through\nextensive empirical analysis and demonstrate that reducing the intra-modal\noverlap leads to a) improved performance on a number of standard datasets, b)\nincreased robustness to distribution shift and c) higher feature variance\nrendering the features more discriminative for downstream tasks.","PeriodicalId":501130,"journal":{"name":"arXiv - CS - Computer Vision and Pattern Recognition","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"CLIP Adaptation by Intra-modal Overlap Reduction\",\"authors\":\"Alexey Kravets, Vinay Namboodiri\",\"doi\":\"arxiv-2409.11338\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Numerous methods have been proposed to adapt a pre-trained foundational CLIP\\nmodel for few-shot classification. As CLIP is trained on a large corpus, it\\ngeneralises well through adaptation to few-shot classification. In this work,\\nwe analyse the intra-modal overlap in image space in terms of embedding\\nrepresentation. Our analysis shows that, due to contrastive learning,\\nembeddings from CLIP model exhibit high cosine similarity distribution overlap\\nin the image space between paired and unpaired examples affecting the\\nperformance of few-shot training-free classification methods which rely on\\nsimilarity in the image space for their predictions. To tackle intra-modal\\noverlap we propose to train a lightweight adapter on a generic set of samples\\nfrom the Google Open Images dataset demonstrating that this improves accuracy\\nfor few-shot training-free classification. We validate our contribution through\\nextensive empirical analysis and demonstrate that reducing the intra-modal\\noverlap leads to a) improved performance on a number of standard datasets, b)\\nincreased robustness to distribution shift and c) higher feature variance\\nrendering the features more discriminative for downstream tasks.\",\"PeriodicalId\":501130,\"journal\":{\"name\":\"arXiv - CS - Computer Vision and Pattern Recognition\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-09-17\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"arXiv - CS - Computer Vision and Pattern Recognition\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/arxiv-2409.11338\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - CS - Computer Vision and Pattern Recognition","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2409.11338","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0

摘要

为了将预先训练好的基础 CLIP 模型适用于少数几次分类,已经提出了许多方法。由于 CLIP 是在大型语料库中训练出来的,因此它能很好地适应少镜头分类。在这项工作中,我们从嵌入表示的角度分析了图像空间中的模内重叠。我们的分析表明,由于对比学习的原因,CLIP 模型的嵌入在配对和非配对实例之间的图像空间中表现出很高的余弦相似性分布重叠,这影响了依赖图像空间相似性进行预测的无训练的少镜头分类方法的性能。为了解决模内重叠问题,我们建议在谷歌开放图片数据集的通用样本集上训练一个轻量级适配器,结果表明这提高了免少量训练分类的准确性。我们通过大量的实证分析验证了我们的贡献,并证明减少模内重叠会带来:a)在一些标准数据集上的性能提高;b)对分布偏移的鲁棒性增强;c)特征变异性提高,使特征对下游任务更具区分性。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
CLIP Adaptation by Intra-modal Overlap Reduction
Numerous methods have been proposed to adapt a pre-trained foundational CLIP model for few-shot classification. As CLIP is trained on a large corpus, it generalises well through adaptation to few-shot classification. In this work, we analyse the intra-modal overlap in image space in terms of embedding representation. Our analysis shows that, due to contrastive learning, embeddings from CLIP model exhibit high cosine similarity distribution overlap in the image space between paired and unpaired examples affecting the performance of few-shot training-free classification methods which rely on similarity in the image space for their predictions. To tackle intra-modal overlap we propose to train a lightweight adapter on a generic set of samples from the Google Open Images dataset demonstrating that this improves accuracy for few-shot training-free classification. We validate our contribution through extensive empirical analysis and demonstrate that reducing the intra-modal overlap leads to a) improved performance on a number of standard datasets, b) increased robustness to distribution shift and c) higher feature variance rendering the features more discriminative for downstream tasks.
求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
自引率
0.00%
发文量
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信