Adverserial_Attack

KO | EN

gitlite — search

#javascript #python #hacktoberfest #ai #golang #go #android #rust #deep-learning #linux #java #windows #awesome #mcp

Adverserial_Attack

(★ 19)

Different Adversarial attack methods implemented in PyTorch on CIFAR-10 Dataset

Adverserial_Attack 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

This repository contains Adversarial Attacks on CIFAR-10 dataset implemented in Pytorch:
  1) Fast Gradient Sign Method (Untargeted)
  2) Iterative Fast Gradient Sign Method (Untargeted)
  3) DeepFool
  
It will include more Adversarial Attacks and Defenses Technique in future as well

*) The CIFAR-10 Network is trained on VGG-16 architecture based on [1] . It reaches a Test accuracy of 88%

Results :  ( Test Accuracy is evaluated after adding adverserial noise to the test data)

1) FGSM method:
              a) Epsilon = 0.1  Test Accuracy = 48.26 %
              b) Epsilon = 0.15 Test Accuracy = 43.26 %
              c) Epsilon = 0.2  Test Accuracy = 40.05 %

2) Iterative FGSM Method:
              a) Epsilon = 0.075
                    i) Iterations = 4   Test Accuracy = 33.06 %
                    ii) Iterations = 10  Test Accuracy = 28.27 %
              b) Epsilon = 0.1
                    i) Iterations = 4    Test Accuracy = 23.52 %
                    ii) Iterations = 10   Test Accuracy = 17.98 %
              c) Epsilon = 0.15
                    i) Iterations = 4     Test Accuracy = 12.7%
                    ii) Iterations = 10   Test Accuract = 6.77 %
                    iii) Iterations = 15  Test Accuracy = 5.93 %

References:

[1] Shuying Liu and Weihong Deng. Very deep convolutional neural network based image classifi- cation using small training 
    sample size. In Pattern Recognition (ACPR), 2015 3rd IAPR Asian Conference on, pages 730–734. IEEE, 2015.
 
[2]  Goodfellow, Ian J, Shlens, Jonathon and Szegedy, Christian. "Explaining and harnessing adversarial examples." 
     arXiv preprint arXiv:1412.6572 (2014): .
     
[3]  S. Moosavi-Dezfooli, A. Fawzi and P. Frossard, "DeepFool: A Simple and Accurate Method to Fool Deep Neural Networks," 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, 2016, pp. 2574-2582.

Adverserial_Attack

Adverserial_Attack 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Adverserial_Attack

Adverserial_Attack 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README