structural-entropy

[UAI 2026 Oral] SeSE: Black-Box Uncertainty Quantification for Large Language Models Based on Structural Information Theory, which aims to detect hallucinated content in LLM-generated text.

A complete-state structural framework investigating how one continuous reality produces observable structure across physical, biological, informational, and measurement domains.

Implementation of "Entropic Flow Network for Bayesian Network Structure Learning," accepted at IEEE ICDM 2026.

Unsupervised social bot detection on multi-relational graphs via structural-entropy community detection