Differentiable Automatic Data Augmentation

Yonggang Li, Guosheng Hu, Yongtao Wang, Timothy Hospedales, Neil M Robertson, Yongxin Yang

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Data augmentation (DA) techniques aim to increase data variability, and thus train deep networks with better generalisation. The pioneering AutoAugment automated the search for optimal DA policies with reinforcement learning. However, AutoAugment is extremely computationally expensive, limiting its wide applicability. Followup work such as PBA and Fast AutoAugment improved efficiency, but optimization speed remains a bottleneck. In this paper, we propose Differentiable Automatic Data Augmentation (DADA) which dramatically reduces the cost. DADA relaxes the discrete DA policy selection to a differentiable optimization problem via Gumbel-Softmax. In addition, we introduce an unbiased gradient estimator, RELAX, leading to an efficient and effective one-pass optimization strategy to learn an efficient and accurate DA policy. We conduct extensive experiments on CIFAR-10, CIFAR-100, SVHN, and ImageNet datasets. Furthermore, we demonstrate the value of Auto DA in pre-training for downstream detection problems. Results show our DADA is at least one order of magnitude faster than the state-of-the-art while achieving very comparable accuracy.
Original languageEnglish
Title of host publicationComputer Vision – ECCV 2020
Subtitle of host publication16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XXII
PublisherSpringer
Pages580-595
Number of pages16
ISBN (Electronic)978-3-030-58542-6
ISBN (Print)978-3-030-58541-9
DOIs
Publication statusPublished - 17 Nov 2020
Event16th European Conference on Computer Vision - Virtual conference
Duration: 23 Aug 202028 Aug 2020
https://eccv2020.eu/

Publication series

NameLecture Notes in Computer Science
PublisherSpringer, Cham
Volume12367
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference16th European Conference on Computer Vision
Abbreviated titleECCV 2020
CityVirtual conference
Period23/08/2028/08/20
Internet address

Keywords / Materials (for Non-textual outputs)

  • AutoML
  • Data Augmentation
  • Differentiable Optimization

Fingerprint

Dive into the research topics of 'Differentiable Automatic Data Augmentation'. Together they form a unique fingerprint.

Cite this