Global Geometry Is Not Enough for Vision Representations

Jiwan Chung, Seon Joo Kim

A common assumption in representation learning is that globally well-distributed embeddings support robust and generalizable representations. This focus has shaped both training ob…