arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1906.08412cs.LGstat.ML
收藏

Data Interpolating Prediction:混淆的另一种解释

Data Interpolating Prediction: Alternative Interpretation of Mixup

The University of Tokyo(东京大学);Preferred Networks(Preferred Networks(首选网络公司))

Takuya Shimada, Shoichiro Yamaguchi, Kohei Hayashi, Sosuke Kobayashi

摘要:

通过混合样本的数据增强(诸如Mixup)已被广泛地典型地用于分类任务。然而,由于用于训练的增强样本和用于测试的原始样本之间的差距,这种策略并不总是有效的。该间隙可能阻止分类器学习最优决策边界并且增加泛化误差。为了克服这个问题,我们提出了一个替代框架称为数据插值预测(DIP)。与常见的数据增强扩充,我们封装的样本混合过程中的假设类的分类器,使训练和测试样本被平等对待。我们推导出的泛化界,并表明DIP有助于降低原来的Rademacher复杂性。此外,我们的经验表明,DIP可以优于现有的Mixup。

英文摘要:

Data augmentation by mixing samples, such as Mixup, has widely been used typically for classification tasks. However, this strategy is not always effective due to the gap between augmented samples for training and original samples for testing. This gap may prevent a classifier from learning the optimal decision boundary and increase the generalization error. To overcome this problem, we propose an alternative framework called Data Interpolating Prediction (DIP). Unlike common data augmentations, we encapsulate the sample-mixing process in the hypothesis class of a classifier so that train and test samples are treated equally. We derive the generalization bound and show that DIP helps to reduce the original Rademacher complexity. Also, we empirically demonstrate that DIP can outperform existing Mixup.

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。

↑