Skip to main content

Provably Learning Diverse Features in Multi-View Data with Midpoint Mixup

Publication ,  Conference
Chidambaram, M; Wang, X; Wu, C; Ge, R
Published in: Proceedings of Machine Learning Research
January 1, 2023

Mixup is a data augmentation technique that relies on training using random convex combinations of data points and their labels. In recent years, Mixup has become a standard primitive used in the training of state-of-the-art image classification models due to its demonstrated benefits over empirical risk minimization with regards to generalization and robustness. In this work, we try to explain some of this success from a feature learning perspective. We focus our attention on classification problems in which each class may have multiple associated features (or views) that can be used to predict the class correctly. Our main theoretical results demonstrate that, for a non-trivial class of data distributions with two features per class, training a 2-layer convolutional network using empirical risk minimization can lead to learning only one feature for almost all classes while training with a specific instantiation of Mixup succeeds in learning both features for every class. We also show empirically that these theoretical insights extend to the practical settings of image benchmarks modified to have multiple features.

Duke Scholars

Published In

Proceedings of Machine Learning Research

EISSN

2640-3498

Publication Date

January 1, 2023

Volume

202

Start / End Page

5563 / 5599
 

Citation

APA
Chicago
ICMJE
MLA
NLM
Chidambaram, M., Wang, X., Wu, C., & Ge, R. (2023). Provably Learning Diverse Features in Multi-View Data with Midpoint Mixup. In Proceedings of Machine Learning Research (Vol. 202, pp. 5563–5599).
Chidambaram, M., X. Wang, C. Wu, and R. Ge. “Provably Learning Diverse Features in Multi-View Data with Midpoint Mixup.” In Proceedings of Machine Learning Research, 202:5563–99, 2023.
Chidambaram M, Wang X, Wu C, Ge R. Provably Learning Diverse Features in Multi-View Data with Midpoint Mixup. In: Proceedings of Machine Learning Research. 2023. p. 5563–99.
Chidambaram, M., et al. “Provably Learning Diverse Features in Multi-View Data with Midpoint Mixup.” Proceedings of Machine Learning Research, vol. 202, 2023, pp. 5563–99.
Chidambaram M, Wang X, Wu C, Ge R. Provably Learning Diverse Features in Multi-View Data with Midpoint Mixup. Proceedings of Machine Learning Research. 2023. p. 5563–5599.

Published In

Proceedings of Machine Learning Research

EISSN

2640-3498

Publication Date

January 1, 2023

Volume

202

Start / End Page

5563 / 5599