arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2002.11102cs.LGcs.CLcs.CVstat.ML
收藏

特征规范化与数据增强

On Feature Normalization and Data Augmentation

Cornell University(康奈尔大学);Cornell Tech(康奈尔科技校区);ASAPP(ASAPP公司);Facebook AI(Facebook人工智能实验室)

Boyi Li, Felix Wu, Ser-Nam Lim, Serge Belongie, Kilian Q. Weinberger

摘要:

时刻(A.K.A.,平均值和标准偏差)在训练图像识别模型时通常作为噪声被去除,以增加稳定性并减少训练时间。然而,在图像生成领域,矩扮演着更加核心的角色。研究表明,实例归一化和位置归一化提取的矩可以粗略地捕捉图像的风格和形状信息。而不是被丢弃,这些时刻是工具的生成过程。在本文中,我们提出了矩交换,一个隐式的数据增强方法,鼓励模型利用矩信息也识别模型。具体来说,我们将一个训练图像的学习特征的矩替换为另一个训练图像的学习特征的矩,并对目标标签进行插值--迫使模型除了从归一化特征中提取训练信号外,还从矩中提取训练信号。由于我们的方法速度快,完全在特征空间中操作,并且混合了与现有方法不同的信号,因此可以将其与现有的增强方法有效地结合起来。我们在几个识别基准数据集上证明了它的有效性,它提高了高度竞争的基线网络的泛化能力,具有显着的一致性。

英文摘要:

The moments (a.k.a., mean and standard deviation) of latent features are often removed as noise when training image recognition models, to increase stability and reduce training time. However, in the field of image generation, the moments play a much more central role. Studies have shown that the moments extracted from instance normalization and positional normalization can roughly capture style and shape information of an image. Instead of being discarded, these moments are instrumental to the generation process. In this paper we propose Moment Exchange, an implicit data augmentation method that encourages the model to utilize the moment information also for recognition models. Specifically, we replace the moments of the learned features of one training image by those of another, and also interpolate the target labels -- forcing the model to extract training signal from the moments in addition to the normalized features. As our approach is fast, operates entirely in feature space, and mixes different signals than prior methods, one can effectively combine it with existing augmentation approaches. We demonstrate its efficacy across several recognition benchmark data sets where it improves the generalization capability of highly competitive baseline networks with remarkable consistency.

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。

↑