Reproduce “NetCause: Counterfactual Learning for Root Cause Analysis in Large-Scale Networks” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Fabien Chraim, Jian Zhang, Dominik Janzing, Xiang Song, Christos Faloutsos, John Evans

Amazon Web Services developed NETCAUSE, a framework for root cause analysis in large-scale networks, which combines a generative spatiotemporal world model with counterfactual simulation to attribute causal influence. Evaluated on production incidents, NETCAUSE achieved an exact match accuracy of 35.5% for top-ranked root causes, an 82% relative improvement over rule-based heuristics.

Key insights

  • Modeling incidents as heterogeneous graph-temporal structures (physical, logical, hierarchical) enables the system to capture intricate dependencies and fault propagation pathways.
  • Utilizing counterfactual reasoning provides a principled, model-based measure of causal influence, moving beyond mere correlation to robustly attribute root causes.
  • Learning fault propagation dynamics directly from incident data in a self-supervised manner allows for adaptivity to evolving network conditions and novel failure modes without extensive manual labeling.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.