Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Fu, Rong, Meng, Chunlei, Liu, Jinshuo, Zhao, Dianyu, Liu, Yongtai, Meng, Yibo, Ma, Xiaowen, Wu, Wangyu, Zeng, Yangchen, Cao, Shuaishuai, Fong, Simon
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2603.01168
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866911613692411904
author Fu, Rong
Meng, Chunlei
Liu, Jinshuo
Zhao, Dianyu
Liu, Yongtai
Meng, Yibo
Ma, Xiaowen
Wu, Wangyu
Zeng, Yangchen
Cao, Shuaishuai
Fong, Simon
author_facet Fu, Rong
Meng, Chunlei
Liu, Jinshuo
Zhao, Dianyu
Liu, Yongtai
Meng, Yibo
Ma, Xiaowen
Wu, Wangyu
Zeng, Yangchen
Cao, Shuaishuai
Fong, Simon
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2603_01168
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle SphUnc: Hyperspherical Uncertainty Decomposition and Causal Identification via Information Geometry
Fu, Rong
Meng, Chunlei
Liu, Jinshuo
Zhao, Dianyu
Liu, Yongtai
Meng, Yibo
Ma, Xiaowen
Wu, Wangyu
Zeng, Yangchen
Cao, Shuaishuai
Fong, Simon
Machine Learning
Artificial Intelligence
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.
title SphUnc: Hyperspherical Uncertainty Decomposition and Causal Identification via Information Geometry
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2603.01168