Data Augmentation for Meta-Learning

Renkun Ni, Micah Goldblum, Amr Sharaf, Kezhi Kong, Tom Goldstein

Keywords: meta-learning, few-shot classification, data augmentation, transfer learning

Abstract: Conventional image classifiers are trained by randomly sampling mini-batches of images. To achieve state-of-the-art performance, sophisticated data augmentation schemes are used to expand the amount of training data available for sampling. In contrast, meta-learning algorithms sample not only images, but classes as well. We investigate how data augmentation can be used not only to expand the number of images available per class, but also to generate entirely new classes. We systematically dissect the meta-learning pipeline and investigate the distinct ways in which data augmentation can be integrated at both the image and class levels. Our proposed meta-specific data augmentation significantly improves the performance of meta-learners on few-shot classification benchmarks.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

One-sentence Summary: We dissect the meta-learning framework and show that meta-specific data augmentation strategies can significantly improve performance.

Community Implementations: [ 1 code implementation](https://www.catalyzex.com/paper/arxiv:2010.07092/code)

Reviewed Version (pdf): https://openreview.net/references/pdf?id=Fc7R7eCcBA