English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Network intrusion detection systems (NIDS) are an essential defense forcomputer networks and the hosts within them. Machine learning (ML) nowadayspredominantly serves as the basis for NIDS decision making, where models aretuned to reduce false alarms, increase detection rates, and detect known andunknown attacks. At the same time, ML models have been found to be vulnerableto adversarial examples that undermine the downstream task. In this work, weask the practical question of whether real-world ML-based NIDS can becircumvented by crafted adversarial flows, and if so, how can they be created.We develop the generative adversarial network (GAN)-based attack algorithmNIDSGAN and evaluate its effectiveness against realistic ML-based NIDS. Twomain challenges arise for generating adversarial network traffic flows: (1) thenetwork features must obey the constraints of the domain (i.e., representrealistic network behavior), and (2) the adversary must learn the decisionbehavior of the target NIDS without knowing its model internals (e.g.,architecture and meta-parameters) and training data. Despite these challenges,the NIDSGAN algorithm generates highly realistic adversarial traffic flows thatevade ML-based NIDS. We evaluate our attack algorithm against twostate-of-the-art DNN-based NIDS in whitebox, blackbox, and restricted-blackboxthreat models and achieve success rates which are on average 99%, 85%, and 70%,respectively. We also show that our attack algorithm can evade NIDS based onclassical ML models including logistic regression, SVM, decision trees andKNNs, with a success rate of 70% on average. Our results demonstrate thatdeploying ML-based NIDS without careful defensive strategies againstadversarial flows may (and arguably likely will) lead to future compromises.

Generating Practical Adversarial Network Traffic Flows Using NIDSGAN