TY  - RPRT
TI  - Beyond Gaussian Bottlenecks: Topologically Aligned Encoding of Vision-Transformer Feature Spaces
AU  - Andrew Bond
AU  - Ilkin Umut Melanlioglu
AU  - Erkut Erdem
AU  - Aykut Erdem
PY  - 2026
UR  - https://arxiv.org/abs/2604.28122
ID  - 2604.28122
ER  -