Packet-Level Adversarial Network Traffic Crafting using Sequence Generative Adversarial Networks

Cheng, Qiumei; Zhou, Shiying; Shen, Yi; Kong, Dezhang; Wu, Chunming

doi:10.48550/arxiv.2103.04794

Cited by 4 publications

(9 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Various researchers have employed GANs to synthesise the network traffic data to address the issue of data imbalance, and adversarial evasion attacks [3], [20], [21]. However, GANs are not good at generating categorical features [8]. Researchers have used one-hot representation [5], or IP2Vec techniques [22], [23] to transform IP addresses into integer values to input to GANs for data generation.…”

Section: F Semantics/functionality Preservationmentioning

confidence: 99%

“…At the same time, the malicious activity must not be compromised. For this reason, several researchers have proposed functionality preservation by modifying only the non-functionality preserving features using GANs [5], [7], [8]. However, various research works claim to preserve the malicious functionality of the generated samples using deep reinforcement learning.…”

Section: F Semantics/functionality Preservationmentioning

confidence: 99%

“…So, GANs do not play a role in generating a complete feature vector in those works. It is also quite challenging to generate categorical features using a GAN without manual engineering except using a sequence GAN [8]. Researchers have also used deep reinforcement learning to generate the functionality preserving adversarial evasion attacks [9]- [11].…”

mentioning

confidence: 99%

See 2 more Smart Citations

Deep Reinforcement Learning based Evasion Generative Adversarial Network for Botnet Detection

Randhawa¹,

Aslam²,

Alauthman³

et al. 2022

Preprint

View full text Add to dashboard Cite

Botnet detectors based on machine learning are potential targets for adversarial evasion attacks. Several research works employ adversarial training with samples generated from generative adversarial nets (GANs) to make the botnet detectors adept at recognising adversarial evasions. However, the synthetic evasions may not follow the original semantics of the input samples. This paper proposes a novel GAN model leveraged with deep reinforcement learning (DRL) to explore semantic aware samples and simultaneously harden its detection. A DRL agent is used to attack the discriminator of the GAN that acts as a botnet detector. The discriminator is trained on the crafted perturbations by the agent during the GAN training, which helps the GAN generator converge earlier than the case without DRL. We name this model RELEVAGAN, i.e. ["relive a GAN" or deep REinforcement Learning-based Evasion Generative Adversarial Network] because, with the help of DRL, it minimises the GAN's job by letting its generator explore the evasion samples within the semantic limits. During the GAN training, the attacks are conducted to adjust the discriminator weights for learning crafted perturbations by the agent. RELEVAGAN does not require adversarial training for the ML classifiers since it can act as an adversarial semantic-aware botnet detection model. Code will be available at https://github.com/rhr407/RELEVAGAN.

show abstract

Section: F Semantics/functionality Preservationmentioning

confidence: 99%

Section: F Semantics/functionality Preservationmentioning

confidence: 99%

See 1 more Smart Citation

Deep Reinforcement Learning based Evasion Generative Adversarial Network for Botnet Detection

Randhawa¹,

Aslam²,

Alauthman³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…The accuracy rates for DT, CNN, and MLP classifiers under the CIFGSM attack dropped to 0.25, 0.68, and 0.73, respectively. Cheng et al [48] proposed Attack-GAN that crafts adversarial traffic at the packet level that maintains the domain constraints. The Attack-GAN utilizes Sequence Generative Adversarial Nets (SeqGAN) with policy gradient in which the generation of adversarial packets is constructed as a sequential decisionmaking process.…”

Section: A Generation Of Aes To Attack Ml-based Nids Modelsmentioning

confidence: 99%

Adversarial Machine Learning In Network Intrusion Detection Domain: A Systematic Review

Alatwi¹,

Morisset²

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…So, GANs do not play a role in generating a complete feature vector in those works. It is also quite challenging to generate categorical features using a GAN without manual engineering except using a sequence GAN [9]. Researchers have also used deep reinforcement learning (DRL) to generate the functionality preserving adversarial evasion attacks [10][11][12].…”

Section: Introductionmentioning

confidence: 99%

Deep Reinforcement Learning Based Evasion Generative Adversarial Network for Botnet Detection

Randhawa

Aslam

Alauthman

et al. 2023

Preprint

View full text Add to dashboard Cite

Packet-Level Adversarial Network Traffic Crafting using Sequence Generative Adversarial Networks

Cited by 4 publications

References 23 publications

Deep Reinforcement Learning based Evasion Generative Adversarial Network for Botnet Detection

Deep Reinforcement Learning based Evasion Generative Adversarial Network for Botnet Detection

Adversarial Machine Learning In Network Intrusion Detection Domain: A Systematic Review

Deep Reinforcement Learning Based Evasion Generative Adversarial Network for Botnet Detection

Contact Info

Product

Resources

About