FairPFL
FairPFL is a federated learning framework for IoT intrusion detection that treats fairness as an operational requirement: no device population should be left systematically under-protected by the shared detector. It combines a self-correcting fairness controller in the aggregation rule, a split personalization architecture, and a selective noise mechanism that perturbs only the shared backbone. This repository contains everything needed to reproduce the experiments. Datasets, trained checkpoints and result directories are not included; the sections below explain how to regenerate them.