- Title
- Robust image classification using a low-pass activation function and DCT augmentation
- Creator
- Hossain, Md Tahmid; Teng, Shyh; Sohel, Ferdous; Lu, Guojun
- Date
- 2021
- Type
- Text; Journal article
- Identifier
- http://researchonline.federation.edu.au/vital/access/HandleResolver/1959.17/178802
- Identifier
- vital:15468
- Identifier
-
https://doi.org/10.1109/ACCESS.2021.3089598
- Identifier
- ISBN:2169-3536 (ISSN)
- Abstract
- Convolutional Neural Network's (CNN's) performance disparity on clean and corrupted datasets has recently come under scrutiny. In this work, we analyse common corruptions in the frequency domain, i.e., High Frequency corruptions (HFc, e.g., noise) and Low Frequency corruptions (LFc, e.g., blur). Although a simple solution to HFc is low-pass filtering, ReLU - a widely used Activation Function (AF), does not have any filtering mechanism. In this work, we instill low-pass filtering into the AF (LP-ReLU) to improve robustness against HFc. To deal with LFc, we complement LP-ReLU with Discrete Cosine Transform based augmentation. LP-ReLU, coupled with DCT augmentation, enables a deep network to tackle the entire spectrum of corruption. We use CIFAR-10-C and Tiny ImageNet-C for evaluation and demonstrate improvements of 5% and 7.3% in accuracy respectively, compared to the State-Of-The-Art (SOTA). We further evaluate our method's stability on a variety of perturbations in CIFAR-10-P and Tiny ImageNet-P, achieving new SOTA in these experiments as well. To further strengthen our understanding regarding CNN's lack of robustness, a decision space visualisation process is proposed and presented in this work. © 2013 IEEE.
- Publisher
- Institute of Electrical and Electronics Engineers Inc.
- Relation
- IEEE Access Vol. 9, no. (2021), p. 86460-86474
- Rights
- All metadata describing materials held in, or linked to, the repository is freely available under a CC0 licence
- Rights
- https://creativecommons.org/licenses/by/4.0/
- Rights
- Copyright @ IEEE
- Rights
- Open Access
- Subject
- 08 Information and Computing Sciences; 09 Engineering; 10 Technology; Activation function; Data augmentation; Image corruption; Input corruption; Low-pass filtering; Robust image classification
- Full Text
- Reviewed
- Hits: 1611
- Visitors: 1633
- Downloads: 100
| Thumbnail | File | Description | Size | Format | |||
|---|---|---|---|---|---|---|---|
| View Details Download | SOURCE1 | Published version | 2 MB | Adobe Acrobat PDF | View Details Download |