SMP-IF

SMP-IF Service Mode Partitioning with Localized Isolation Forests for Unsupervised Network Flow Anomaly Detection This repository contains the implementation and experimental artifacts for SMP-IF, an anomaly detection framework that partitions network traffic into service-specific modes and trains localized Isolation Forest models for each mode. The evaluation uses the CIC-IDS-2017 dataset. Models are trained on benign Monday traffic targeting host 192.168.10.50 and evaluated on Tuesday–Friday traffic containing both benign and attack flows.