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 1811.07484cs.CVcs.LG
收藏

锐化焦点:注意力分离和一致性的学习

Sharpen Focus: Learning with Attention Separability and Consistency

Rutgers University(罗格斯大学);Siemens Corporate Technology(西门子企业技术)

Lezi Wang, Ziyan Wu, Srikrishna Karanam, Kuan-Chuan Peng, Rajat Vikram Singh, Bo Liu, Dimitris N. Metaxas

摘要:

基于梯度的注意力建模的最新发展已经看到注意力地图成为解释卷积神经网络的强大工具。尽管对于感兴趣的单个类别具有良好的定位,但这些技术产生的注意力图在不同类别之间具有实质上重叠的响应,从而导致视觉混乱的问题和对区分性注意力的需要。在本文中,我们解决这个问题的一个新的框架,使类歧视性的注意学习过程中的原则性的一部分。我们的关键创新包括针对注意力可分离性和跨层一致性的新学习目标,从而提高注意力可区分性并减少视觉混乱。在图像分类基准上的大量实验表明,我们的方法在提高分类准确率方面是有效的,包括CIFAR-100(+3.33%),Caltech-256(+1.64%),ILSVRC 2012(+0.92%),CUB-200-2011(+4.8%)和PASCAL VOC 2012(+5.73%)。

英文摘要:

Recent developments in gradient-based attention modeling have seen attention maps emerge as a powerful tool for interpreting convolutional neural networks. Despite good localization for an individual class of interest, these techniques produce attention maps with substantially overlapping responses among different classes, leading to the problem of visual confusion and the need for discriminative attention. In this paper, we address this problem by means of a new framework that makes class-discriminative attention a principled part of the learning process. Our key innovations include new learning objectives for attention separability and cross-layer consistency, which result in improved attention discriminability and reduced visual confusion. Extensive experiments on image classification benchmarks show the effectiveness of our approach in terms of improved classification accuracy, including CIFAR-100 (+3.33%), Caltech-256 (+1.64%), ILSVRC2012 (+0.92%), CUB-200-2011 (+4.8%) and PASCAL VOC2012 (+5.73%).

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。

↑