Deep Reinforcement Adversarial Learning Against Botnet Evasion Attacks

Apruzzese, Giovanni; Andreolini, Mauro; Marchetti, Mirco; Venturi, Andrea; Colajanni, Michele

doi:10.1109/tnsm.2020.3031843

Cited by 73 publications

(56 citation statements)

References 62 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As shown in the related paper [6] , the generated adversarial samples are able of evading not only the RF detector used for training the DRL agents, but also the WnD detector. In the dataset we include only those perturbed samples that are able of evading the detection with less than 80 actions.…”

Section: Experimental Design Materials and Methodsmentioning

confidence: 91%

“…To implement and train the classifiers we use the Scikit-learn framework [11] (version 0.21.2). The provided DReLAB dataset allows the implementation of classifiers with high detection rates, achieving Recall scores often superior to 0.95 (refer to the primary research paper [6] for more information).…”

Section: Experimental Design Materials and Methodsmentioning

confidence: 99%

“…The workflow for training a DRL agent to produce adversarial samples related to a botnet family b proceeds as follows (we refer to Figure 4 of the related research paper [6] for an illustration of the process). Each malicious flow of b, which we denote f b , is processed and modified individually.…”

Section: Experimental Design Materials and Methodsmentioning

confidence: 99%

“…“Deep Reinforcement Adversarial Learning against Botnet Evasion Attacks.” IEEE Transactions on Network and Service Management . DOI: 10.1109/TNSM.2020.3031843 [6] …”

Section: Specifications Tablementioning

confidence: 99%

“… Our dataset al1lows researchers to focus only on the proposal of novel defensive strategies without working on procedures for generating meaningful adversarial samples. Furthermore, the proposed dataset includes samples from a novel source, CSE-CIC-IDS2018 [3] , that is not considered in the main research paper [6] . This additional source further enhances the quality of the proposed dataset as it contains thousands of adversarial samples belonging to several botnet families and resembling multiple attack scenarios.…”

Section: Value Of the Datamentioning

confidence: 99%

See 4 more Smart Citations

DReLAB - Deep REinforcement Learning Adversarial Botnet: A benchmark dataset for adversarial attacks against botnet Intrusion Detection Systems

et al. 2021

Self Cite

View full text Add to dashboard Cite

We present the first dataset that aims to serve as a benchmark to validate the resilience of botnet detectors against adversarial attacks. This dataset includes realistic adversarial samples that are generated by leveraging two widely used Deep Reinforcement Learning (DRL) techniques. These adversarial samples are proved to evade state of the art detectors based on Machine- and Deep-Learning algorithms. The initial corpus of malicious samples consists of network flows belonging to different botnet families presented in three public datasets containing real enterprise network traffic. We use these datasets to devise detectors capable of achieving state-of-the-art performance. We then train two DRL agents, based on Double Deep Q-Network and Deep Sarsa , to generate realistic adversarial samples: the goal is achieving misclassifications by performing small modifications to the initial malicious samples. These alterations involve the features that can be more realistically altered by an expert attacker, and do not compromise the underlying malicious logic of the original samples. Our dataset represents an important contribution to the cybersecurity research community as it is the first including thousands of automatically generated adversarial samples that are able to thwart state of the art classifiers with a high evasion rate. The adversarial samples are grouped by malware variant and provided in a CSV file format. Researchers can validate their defensive proposals by testing their detectors against the adversarial samples of the proposed dataset. Moreover, the analysis of these samples can pave the way to a deeper comprehension of adversarial attacks and to some sort of explainability of machine learning defensive algorithms. They can also support the definition of novel effective defensive techniques.

show abstract

Section: Experimental Design Materials and Methodsmentioning

confidence: 91%

Section: Experimental Design Materials and Methodsmentioning

confidence: 99%

Section: Experimental Design Materials and Methodsmentioning

confidence: 99%

“…“Deep Reinforcement Adversarial Learning against Botnet Evasion Attacks.” IEEE Transactions on Network and Service Management . DOI: 10.1109/TNSM.2020.3031843 [6] …”

Section: Specifications Tablementioning

confidence: 99%