TY  - RPRT
TI  - Can multimodal representation learning by alignment preserve modality-specific information?
AU  - Romain Thoreau
AU  - Jessie Levillain
AU  - Dawa Derksen
PY  - 2025
UR  - https://arxiv.org/abs/2509.17943
ID  - 2509.17943
ER  -