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 language | English |
---|---|
Title of host publication | Computer Vision – ECCV 2020 |
Subtitle of host publication | 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XXII |
Publisher | Springer |
Pages | 580-595 |
Number of pages | 16 |
ISBN (Electronic) | 978-3-030-58542-6 |
ISBN (Print) | 978-3-030-58541-9 |
DOIs | |
Publication status | Published - 17 Nov 2020 |
Event | 16th European Conference on Computer Vision - Virtual conference Duration: 23 Aug 2020 → 28 Aug 2020 https://eccv2020.eu/ |
Publication series
Name | Lecture Notes in Computer Science |
---|---|
Publisher | Springer, Cham |
Volume | 12367 |
ISSN (Print) | 0302-9743 |
ISSN (Electronic) | 1611-3349 |
Conference
Conference | 16th European Conference on Computer Vision |
---|---|
Abbreviated title | ECCV 2020 |
City | Virtual conference |
Period | 23/08/20 → 28/08/20 |
Internet address |
Keywords / Materials (for Non-textual outputs)
- AutoML
- Data Augmentation
- Differentiable Optimization