mmamun1 / LMTracker

The repository implement the LMTracker model based on paper: LMTracker: Lateral movement path detection based on heterogeneous graph embedding

1. Contruct Graph from LANL data: python3 graph_construction.py

2. Train the metapath2vec model: python3 metapath2vec_torchgeo/metapath2vec.py

3. run autoecoder.ipynb to train from the attack path data generated from metapath2vec

Phase 1: replicate the paper [DONE]

• all snapshot in phase1 folder
• graph folder: graph_data_20220219110042
• path embedding folder: model_20220219164849

Phase 2: expand the graph with the consideration of the authentication type

• in progress