Generalizable Model for Network Security

Overview

Cybersecurity defense mechanisms are crucial for protecting sensitive information within computer networks. However, previous approaches for Intrusion Detection Systems (IDS) fail to address attacks within unique network environments since current datasets represent homogenous networks. In this paper, we curate a generalizable network dataset for intrusion detection that reflects various network conditions. We accomplish this goal by utilizing realistic background traffic based on University of California, Santa Barbara gateway data. Subsequently, we develop a pipeline to run attacks alongside the replayed subnet traffic, shape the traffic bandwidth, and capture relevant data for the proposed dataset. Lastly, we construct our dataset from collected traffic and assess the curated dataset on a proposed IDS dataset evaluation framework. The proposed dataset fulfills 13 of the 14 evaluation criteria—displaying our work’s applicability for future generalizable IDS dataset creation and model development.

Generalizable Model for Network Security

Overview

Competition history

Related projects

Cyber Smart Security: Artificial Intelligence for Network Intrusion Detection Systems

Hybrid CNN-LSTM Network Intrusion Detection System Utilizing Multi-Task Learning and Simulated Network Defense

FKAN-Enhanced Multi-View Self-Supervised Network Intrusion Detection with Frequency-Aware Temporal Modeling

Using Machine Learning to Detect Computer Network Security Threats

Building a Deep Neural Network to Automate Protection Against Online Cyber Attacks

A Deep-Learning Based Cascading Framework for Intrusion Detection and Attack Classification on V2X-Based Autonomous Vehi

Towards Malware Classifiers Robust to Adversarial Malware

Cyber Automated Report Linker: A Network Approach to Minimizing Expansion of Catastrophic Cyber Infiltrations

Browse more like this