arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2501.02860cs.LGcs.CV
收藏

在自监督表征学习中从局部看见整体

Seeing the Whole in the Parts in Self-Supervised Representation Learning

Frankfurt Institute for Advanced Studies(法兰克福高级研究院);Institut Pascal(帕斯卡尔研究院);CNRS(法国国家科学研究中心);Clermont Auvergne INP(克莱蒙奥弗涅国立综合理工学院)

Arthur Aubret, Céline Teulière, Jochen Triesch

AI总结:

本文提出 CO-SSL,通过对齐池化前的局部表征与全局图像表征建模空间共现,在 ImageNet-1K 等数据集上取得更优准确率与鲁棒性。

摘要:

自监督学习(SSL)近期的成功通过遮蔽图像部分区域或对其进行激进裁剪,来对视觉特征的空间共现进行建模。本文提出一种建模空间共现的新方法:将局部表征(池化之前)与全局图像表征对齐。我们提出 CO-SSL,这是一族实例判别方法,并证明其在多个数据集上优于以往方法,包括 ImageNet-1K:在该数据集上经过100个预训练 epoch,其 Top-1 准确率达到71.5%。CO-SSL 对噪声损坏、内部损坏、小规模对抗攻击以及较大的训练裁剪尺寸也更加鲁棒。我们的分析进一步表明,CO-SSL 学到了高度冗余的局部表征,这为其鲁棒性提供了解释。总体而言,本研究表明,对齐局部与全局表征可能是无监督类别学习的一条强有力原则。

英文摘要:

Recent successes in self-supervised learning (SSL) model spatial co-occurrences of visual features either by masking portions of an image or by aggressively cropping it. Here, we propose a new way to model spatial co-occurrences by aligning local representations (before pooling) with a global image representation. We present CO-SSL, a family of instance discrimination methods and show that it outperforms previous methods on several datasets, including ImageNet-1K where it achieves 71.5% of Top-1 accuracy with 100 pre-training epochs. CO-SSL is also more robust to noise corruption, internal corruption, small adversarial attacks, and large training crop sizes. Our analysis further indicates that CO-SSL learns highly redundant local representations, which offers an explanation for its robustness. Overall, our work suggests that aligning local and global representations may be a powerful principle of unsupervised category learning.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑