DOT-CBM uses optimal transport between image patches and concept embeddings, with disentanglement and bias priors, to improve accuracy and localize concepts.
Sinkhorn distances: Lightspeed computation of optimal transport
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Discovering Fine-Grained Visual-Concept Relations by Disentangled Optimal Transport Concept Bottleneck Models
DOT-CBM uses optimal transport between image patches and concept embeddings, with disentanglement and bias priors, to improve accuracy and localize concepts.