SphUnc: Hyperspherical Uncertainty Decomposition and Causal Identification via Information Geometry

2026-03-01

AI Summary

This work addresses the lack of calibratable and interpretable uncertainty modeling coupled with causal reasoning in multi-agent systems. The authors propose a novel approach that integrates information geometry with structural causal models by mapping features to von Mises–Fisher distributions on a unit hypersphere. This representation enables a principled decomposition of epistemic and aleatoric uncertainty through information-geometric analysis, while a causal graph constructed over the spherical latent variables supports interventional reasoning. To the best of our knowledge, this is the first method to achieve disentangled uncertainty quantification and interpretable causal identification under high-order interactions within a hyperspherical representation space. Experiments demonstrate significant improvements in both predictive accuracy and calibration on social and affective benchmarks, alongside the provision of interpretable causal signals.

Technology Category

• Reasoning under Uncertainty: Causality
• Machine Learning: Calibration & Uncertainty Quantification
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Reliable decision-making in complex multi-agent systems requires calibrated predictions and interpretable uncertainty. We introduce SphUnc, a unified framework combining hyperspherical representation learning with structural causal modeling. The model maps features to unit hypersphere latents using von Mises-Fisher distributions, decomposing uncertainty into epistemic and aleatoric components through information-geometric fusion. A structural causal model on spherical latents enables directed influence identification and interventional reasoning via sample-based simulation. Empirical evaluations on social and affective benchmarks demonstrate improved accuracy, better calibration, and interpretable causal signals, establishing a geometric-causal foundation for uncertainty-aware reasoning in multi-agent settings with higher-order interactions.

Problem

• multi-agent systems
• uncertainty decomposition
• calibrated predictions
• interpretable uncertainty
• causal identification

Innovation

• hyperspherical representation
• uncertainty decomposition
• information geometry
• structural causal model
• multi-agent systems