An Analysis of Transferability in Network Intrusion Detection using Distributed Deep Learning

Shreya Ghosh, Abu Shafin Mohammad Mahdee Jameel, Aly El Gamal

Abstract: In this paper, we utilize a distributed deep learning framework to investigate transferability of network intrusion detection between federated nodes. Transferable learning makes intrusion detection systems more robust to rare attacks and enables them to adapt to real life scenarios. We analyze symmetric and asymmetric transferability relationships. We propose and investigate the impact of feature pre-processing to improve transferability. The code for this work is available at https://github.com/ghosh64/fedlearn.