TY  - RPRT
TI  - Contrastive Learning Can Find An Optimal Basis For Approximately View-Invariant Functions
AU  - Daniel D. Johnson
AU  - Ayoub El Hanchi
AU  - Chris J. Maddison
PY  - 2023
UR  - https://arxiv.org/abs/2210.01883
ID  - 2210.01883
ER  -