Generating Adversarial Malware Examples for Black-Box Attacks Based on GAN (2017-02-20T00:00:00.000000Z)

TL;DR

The superiority of MalGAN over traditional gradient based adversarial example generation algorithms is that MalGAN is able to decrease the detection rate to nearly zero and make the retraining based defensive method against adversarial examples hard to work.

Abstract

Machine learning has been used to detect new malware in recent years, while malware authors have strong motivation to attack such algorithms. Malware authors usually have no access to the detailed structures and parameters of the machine learning models used by malware detection systems, and therefore they can only perform black-box attacks. This paper proposes a generative adversarial network (GAN) based algorithm named MalGAN to generate adversarial malware examples, which are able to bypass black-box machine learning based detection models. MalGAN uses a substitute detector to fit the black-box malware detection system. A generative network is trained to minimize the generated adversarial examples' malicious probabilities predicted by the substitute detector. The superiority of MalGAN over traditional gradient based adversarial example generation algorithms is that MalGAN is able to decrease the detection rate to nearly zero and make the retraining based defensive method against adversarial examples hard to work.

Authors

Weiwei Hu

1 papers

Ying Tan

1 papers

Generating Adversarial Malware Examples for Black-Box Attacks Based on GAN

TL;DR

Abstract

Authors

References29 items

Evaluation of Defensive Methods for DNNs against Multiple Adversarial Evasion Models

Towards Principled Methods for Training Generative Adversarial Networks

Delving into Transferable Adversarial Examples and Black-box Attacks

Adversarial Perturbations Against Deep Neural Networks for Malware Classification

Improved Techniques for Training GANs

Transferability in Machine Learning: from Phenomena to Black-Box Attacks using Adversarial Samples

A General Retraining Framework for Scalable Adversarial Classification

Practical Black-Box Attacks against Machine Learning

Practical Black-Box Attacks against Deep Learning Systems using Adversarial Examples

The Limitations of Deep Learning in Adversarial Settings

Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks

Distillation as a Defense to Adversarial Perturbations Against Deep Neural Networks

Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks

Adam: A Method for Stochastic Optimization

Explaining and Harnessing Adversarial Examples

Towards Deep Neural Network Architectures Robust to Adversarial Examples

Conditional Generative Adversarial Nets

Intriguing properties of neural networks

Et al

Learning to Detect and Classify Malicious Executables in the Wild

Learning to detect malicious executables in the wild

Data mining methods for detection of new malicious executables

GENERATIVE ADVERSARIAL NETS

In NIPS 2016 Workshop on Adversarial Training

and Yevgeniy Vorobeychik

Microsoft, 2013] Microsoft. Microsoft portable executable and common object file format specification

The Journal of Machine Learning Research

pages 470–478

In Advances in Neural Information Processing Systems

Field of Study

Venue Information

Name

Type

URL

Alternate Names