Logo of Takinur
Network Intrusion Detection System (NIDS) using Few-Shot Learning

Network Intrusion Detection System (NIDS) using Few-Shot Learning

Tagline

Few-shot anomaly detection for zero-day attacks.

Tech Stack

Python , TensorFlow/PyTorch , Scikit-learn , Pandas

Year
2025

Developed an anomaly detection model using Few-Shot Learning and Siamese Networks to identify zero-day network attacks with limited training data. Achieved 94% detection accuracy on 5 unseen attack classes, outperforming a supervised XGBoost baseline by 11pp in 5-fold cross-validation, and demonstrated clear separation between benign traffic and malicious intrusion patterns.

Glimpse of the Project

Focused on model development, evaluation, and intrusion pattern separation rather than UI screenshots.

next project
Original Paintings