Explainable AI for Jet Tagging: A Comparative Study of GNNExplainer, GNNShap, and GradCAM for Jet Tagging in the Lund Jet Plane

2026-04-28

AI Summary

Graph neural networks (GNNs) have demonstrated strong performance in jet tagging, yet their decisions lack interpretability, raising concerns about whether they truly learn underlying physical mechanisms. This work adapts GNNExplainer, GNNShap, and GradCAM to the Lund jet plane representation and constructs physics-informed explanation masks using Monte Carlo truth labels. It further introduces the first interpretability evaluation framework tailored to jet physics, which systematically assesses the correlation between various explanation methods and canonical jet substructure observables—such as N-subjettiness ratios and energy correlation functions—across different transverse momentum regimes. The results reveal that node importance scores exhibit significant correlations with analytical observables and evolve consistently across phase space, confirming that GNNs indeed capture key physical features of jet substructure.

Abstract

Graph neural networks such as ParticleNet and transformer based networks on point clouds such as ParticleTransformer achieve state-of-the-art performance on jet tagging benchmarks at the Large Hadron Collider, yet the physical reasoning behind their predictions remains opaque. We present different methods, i.e. perturbation-based (GNNExplainer), Shapley-value-based (GNNShap), and gradient-based (GRADCam); adapted to operate on LundNet's Lund-plane graph representation. Leveraging the fact that each node in the Lund plane corresponds to a physically meaningful parton splitting, we construct Monte Carlo truth explanation masks and introduce a physics-informed evaluation framework that goes beyond standard fidelity metrics. We perform the analysis in three transverse-momentum bins ($\mathrm{p_T} \in [500,700]$, $[800,1000]$, and the inclusive region $[500,1000]$ GeV), revealing how explanation quality and focus shift between non-perturbative and perturbative regimes. We further quantify the correlation between explainer-assigned node importance and classical jet substructure observables -- $N$-subjettiness ratios $τ_{21}$ and $τ_{32}$ and the energy correlation functions -- establishing the degree to which the model has learned known QCD features. We find that overall the weight assigned by explainability methods has a correlation with analytic observables, with expected shift across different phase space regimes, indicating that a trained neural network indeed learns some aspects of jet-substructure moments. Our open-source implementation enables reproducible explainability studies for graph-based jet taggers.

Problem

• Explainable AI
• Jet Tagging
• Graph Neural Networks
• Lund Jet Plane
• Physics Interpretability

Innovation

• Explainable AI
• Jet Tagging
• Graph Neural Networks
• Lund Jet Plane
• Physics-informed Evaluation