Diagnosing Generalization Failures from Representational Geometry Markers

Published by

cnchou.github.io

on •1 min read

Diagnosing Generalization Failures from Representational Geometry Markers

Generalization—the ability to perform well beyond the training context—is a hallmark of biological and artificial intelligence, yet anticipating unseen failures remains a central challenge.

Representational Geometry Out-Of-Distribution Generalization Neural Manifolds Manifold Capacity AI interpretability Transfer Learning ImageNet Model Selection ICLR 2026

Learn more