Explainable Deep-Learning Based Potentially Hazardous Asteroids Classification Using Graph Neural Networks

TL;DR

This work addresses the hazard classification of potentially hazardous asteroids (PHAs) by modeling the asteroid population as a graph and applying a Graph Neural Network (GNN) with attention to capture dynamical relationships. The pipeline includes SMOTE-based balancing, median imputation, and $k$-NN graph construction ($k=5$), followed by a three-layer GNN that outputs hazard probabilities via a sigmoid head. The approach achieves an overall accuracy of $0.991$ and an AUC of $0.99$, with hazardous-class metrics showing precision $\approx 0.24$, recall $\approx 0.78$, and F1 $\approx 0.37$ after balancing, highlighting a favorable recall for defense needs but a trade-off with false positives. Feature importance points to albedo, perihelion distance $q$, and semi-major axis $a$ as key predictors, enabling explainable hazard reasoning and informing future autonomous navigation and planetary defense applications using missions like NEO Surveyor and Ramses.

Abstract

Classifying potentially hazardous asteroids (PHAs) is crucial for planetary defense and deep space navigation, yet traditional methods often overlook the dynamical relationships among asteroids. We introduce a Graph Neural Network (GNN) approach that models asteroids as nodes with orbital and physical features, connected by edges representing their similarities, using a NASA dataset of 958,524 records. Despite an extreme class imbalance with only 0.22% of the dataset with the hazardous label, our model achieves an overall accuracy of 99% and an AUC of 0.99, with a recall of 78% and an F1-score of 37% for hazardous asteroids after applying the Synthetic Minority Oversampling Technique. Feature importance analysis highlights albedo, perihelion distance, and semi-major axis as main predictors. This framework supports planetary defense missions and confirms AI's potential in enabling autonomous navigation for future missions such as NASA's NEO Surveyor and ESA's Ramses, offering an interpretable and scalable solution for asteroid hazard assessment.

Figures (9)

• Figure 1: Methodology Pipeline
• Figure 2: Feature Distributions
• Figure 3: Class Distribution Before SMOTE
• Figure 4: Class Distribution After SMOTE
• Figure 5: GNN Architecture