← Search

Andrew Tan

2 accepted papers

2026

From Flat Facts to Sharp Hallucinations: Detecting Stubborn Errors via Gradient Sensitivity

ICML 2026poster

Traditional hallucination detection fails on "Stubborn Hallucinations"—errors where LLMs are confidently wrong. We propose a geometric solution: Embedding-Perturbed Gradient Sensitivity (EPGS). We hypothesize that while robust facts reside in flat minima, stubborn hallucinations sit in sharp minima,…

Cited by 0SourceScholar
2020

AI Feynman 2.0: Pareto-optimal symbolic regression exploiting graph modularity

NeurIPS 2020oral

We present an improved method for symbolic regression that seeks to fit data to formulas that are Pareto-optimal, in the sense of having the best accuracy for a given complexity. It improves on the previous state-of-the-art by typically being orders of magnitude more robust toward noise and bad data…

Cited by 272SourcePDFScholar