Timezone: »

 
Spotlight
Data Augmentation as Feature Manipulation
Ruoqi Shen · Sebastien Bubeck · Suriya Gunasekar

Wed Jul 20 07:50 AM -- 07:55 AM (PDT) @ Room 309
in DL: Theory »

Data augmentation is a cornerstone of the machine learning pipeline, yet its theoretical underpinnings remain unclear. Is it merely a way to artificially augment the data set size? Or is it about encouraging the model to satisfy certain invariances? In this work we consider another angle, and we study the effect of data augmentation on the dynamic of the learning process. We find that data augmentation can alter the relative importance of various features, effectively making certain informative but hard to learn features more likely to be captured in the learning process. Importantly, we show that this effect is more pronounced for non-linear models, such as neural networks. Our main contribution is a detailed analysis of data augmentation on the learning dynamic for a two layer convolutional neural network in the recently proposed multi-view model by Z. Allen-Zhu and Y. Li. We complement this analysis with further experimental evidence that data augmentation can be viewed as a form of feature manipulation.

Keywords: DL: Other Representation Learning · DL: Theory · T: Deep Learning · T: Learning Theory

Author Information

Ruoqi Shen (University of Washington)
Sebastien Bubeck (Microsoft Research)
Suriya Gunasekar (Microsoft Research)

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors