https://github.com/mdbloice/patch-augmentation

Code for the implemenation of the Patch Augmentation technique
https://github.com/mdbloice/patch-augmentation

image-augmentation machine-learning

Last synced: 11 months ago
JSON representation

Code for the implemenation of the Patch Augmentation technique

• Host: GitHub
• URL: https://github.com/mdbloice/patch-augmentation
• Owner: mdbloice
• License: mit
• Created: 2019-10-08T14:02:15.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2019-11-28T15:02:24.000Z (over 6 years ago)
• Last Synced: 2025-04-05T16:36:41.171Z (about 1 year ago)
• Topics: image-augmentation, machine-learning
• Language: Jupyter Notebook
• Homepage:
• Size: 18 MB
• Stars: 11
• Watchers: 5
• Forks: 4
• Open Issues: 1
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# Patch Augmentation

*Patch Augmentation* is an novel image augmentation technique designed to improve model generalisation and mitigate against adversarial attacks.

For details, see the following pre-print: **Patch augmentation: Towards efficient decision boundaries for neural networks**, *arXiv:1911.07922*, Nov. 2019, 

## How it works

*Patch Augmentation* is a data-independent approach that creates new image data based on image/label pairs, where a patch from one of the two images in the pair is superimposed on to the other image, creating a new augmented sample. 

Below is a visual example of the technique:

![examples-random-patches-cropped.jpg](./DemoImages/examples-random-patches-cropped.jpg)

The augmented image label is a combination of the image pair's original labels. The labels for the dog and cat classes are `[1.0, 0.0]` and `[0.0, 1.0]` respectively. Clockwise from the upper left, the augmented image's labels are `[0.72220625, 0.27779375]`, `[0.2832, 0.7168]`, `[0.0, 1.0]`, and `[0.918925, 0.081075]` respectively.

A notebook containing a reproducible experiment (training ResNet20v1 using the CIFAR-100 data set) can be found in the following notebook:

[Patch-Augmentation-CIFAR-100.ipynb](Patch-Augmentation-CIFAR-100.ipynb)

In the notebook above, *Patch Augmentation* improves a baseline accuracy of about 45% to over 61%.

The table below shows the technique being applied to several data sets and network architectures (test set accuracy):

| Dataset       | Model      | No Augmentation | Patch Augmentation  |

|---------------|------------|-----------------|---------------------|

| CIFAR-10      | ResNet20v1 | 80.86%          | **89.33%**          |

|               | ResNet29v2 | 83.15%          | **91.19%**          |

| CIFAR-100     | ResNet20v1 | 44.08%          | **61.41%**          |

|               | ResNet29v2 | 52.21%          | **68.06%**          |

## Robustness Against Adversarial Attacks

Initial experiments show networks trained with *Patch Augmentation* are more robust to adversarial attacks, see the following notebook for details:

[Adversarial-Examples.ipynb](Adversarial-Examples.ipynb)

Using the Fast Gradient Sign Method to create adversarial examples, the network trained with *Patch Augmentation* had an accuracy of 72.5% versus 64.3% compared to the network trained without augmentation.

## Publication

Publication in review.

Pre-print available here: **Patch augmentation: Towards efficient decision boundaries for neural networks**, *arXiv:1911.07922*, Nov. 2019, 

Repository made public on the 25th of October 2019.