CNNs outperform Random Forests in robust network intrusion detection

A new research paper evaluates the robustness of machine learning models used in network intrusion detection systems against adversarial attacks. The study tested three popular architectures—1D Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), and Random Forest (RF)—using the ACI-IoT-2023 dataset and FGSM/PGD adversarial attacks. Surprisingly, the Random Forest model, despite high baseline accuracy, collapsed significantly under attack, while the CNN maintained strong performance even with increasing perturbation levels. AI

IMPACT CNN-based architectures are recommended for intrusion detection systems facing adversarial threats, challenging prior assumptions about model performance.

RANK_REASON Academic paper evaluating machine learning model robustness. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• 1D Convolutional Neural Network
• ACI-IoT-2023 dataset
• FGSM
• Long Short-Term Memory
• Random Forest

• paper
• safety

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →