Matches in SemOpenAlex for { <https://semopenalex.org/work/W4312533318> ?p ?o ?g. }
- W4312533318 abstract "Metric learning aims to learn a highly discriminative model encouraging the embeddings of similar classes to be close in the chosen metrics and pushed apart for dissimilar ones. The common recipe is to use an encoder to extract embeddings and a distance-based loss function to match the representations - usually, the Euclidean distance is utilized. An emerging interest in learning hyperbolic data embeddings suggests that hyperbolic geometry can be beneficial for natural data. Following this line of work, we propose a new hyperbolic-based model for metric learning. At the core of our method is a vision transformer with output embeddings mapped to hyperbolic space. These embeddings are directly optimized using modified pairwise cross-entropy loss. We evaluate the proposed model with six different formulations on four datasets achieving the new state-of-the-art performance. The source code is available at https://github.com/htdt/hyp_metric." @default.
- W4312533318 created "2023-01-05" @default.
- W4312533318 creator A5004111307 @default.
- W4312533318 creator A5013359809 @default.
- W4312533318 creator A5027171279 @default.
- W4312533318 creator A5040956144 @default.
- W4312533318 creator A5043761317 @default.
- W4312533318 date "2022-06-01" @default.
- W4312533318 modified "2023-09-30" @default.
- W4312533318 title "Hyperbolic Vision Transformers: Combining Improvements in Metric Learning" @default.
- W4312533318 cites W2001070843 @default.
- W4312533318 cites W2097117768 @default.
- W4312533318 cites W2108598243 @default.
- W4312533318 cites W2115857747 @default.
- W4312533318 cites W2117539524 @default.
- W4312533318 cites W2138011018 @default.
- W4312533318 cites W2138621090 @default.
- W4312533318 cites W2194775991 @default.
- W4312533318 cites W2342045095 @default.
- W4312533318 cites W2471768434 @default.
- W4312533318 cites W2605102252 @default.
- W4312533318 cites W2606377603 @default.
- W4312533318 cites W2889326414 @default.
- W4312533318 cites W2948638722 @default.
- W4312533318 cites W2953271441 @default.
- W4312533318 cites W2963026686 @default.
- W4312533318 cites W2963113119 @default.
- W4312533318 cites W2963350250 @default.
- W4312533318 cites W2963466847 @default.
- W4312533318 cites W2963574614 @default.
- W4312533318 cites W2964105864 @default.
- W4312533318 cites W2964271799 @default.
- W4312533318 cites W2988501586 @default.
- W4312533318 cites W2991234496 @default.
- W4312533318 cites W2991581349 @default.
- W4312533318 cites W3034202663 @default.
- W4312533318 cites W3035014997 @default.
- W4312533318 cites W3035102141 @default.
- W4312533318 cites W3035524453 @default.
- W4312533318 cites W3035700349 @default.
- W4312533318 cites W3098656025 @default.
- W4312533318 cites W3099206234 @default.
- W4312533318 cites W3159481202 @default.
- W4312533318 cites W3211144631 @default.
- W4312533318 cites W4214764301 @default.
- W4312533318 doi "https://doi.org/10.1109/cvpr52688.2022.00726" @default.
- W4312533318 hasPublicationYear "2022" @default.
- W4312533318 type Work @default.
- W4312533318 citedByCount "12" @default.
- W4312533318 countsByYear W43125333182023 @default.
- W4312533318 crossrefType "proceedings-article" @default.
- W4312533318 hasAuthorship W4312533318A5004111307 @default.
- W4312533318 hasAuthorship W4312533318A5013359809 @default.
- W4312533318 hasAuthorship W4312533318A5027171279 @default.
- W4312533318 hasAuthorship W4312533318A5040956144 @default.
- W4312533318 hasAuthorship W4312533318A5043761317 @default.
- W4312533318 hasBestOaLocation W43125333182 @default.
- W4312533318 hasConcept C106301342 @default.
- W4312533318 hasConcept C111919701 @default.
- W4312533318 hasConcept C11413529 @default.
- W4312533318 hasConcept C118505674 @default.
- W4312533318 hasConcept C118615104 @default.
- W4312533318 hasConcept C120174047 @default.
- W4312533318 hasConcept C121332964 @default.
- W4312533318 hasConcept C129782007 @default.
- W4312533318 hasConcept C134306372 @default.
- W4312533318 hasConcept C154945302 @default.
- W4312533318 hasConcept C162324750 @default.
- W4312533318 hasConcept C165801399 @default.
- W4312533318 hasConcept C176217482 @default.
- W4312533318 hasConcept C184898388 @default.
- W4312533318 hasConcept C198043062 @default.
- W4312533318 hasConcept C21547014 @default.
- W4312533318 hasConcept C2524010 @default.
- W4312533318 hasConcept C33923547 @default.
- W4312533318 hasConcept C41008148 @default.
- W4312533318 hasConcept C62520636 @default.
- W4312533318 hasConcept C66322947 @default.
- W4312533318 hasConcept C80444323 @default.
- W4312533318 hasConcept C92047909 @default.
- W4312533318 hasConcept C97931131 @default.
- W4312533318 hasConceptScore W4312533318C106301342 @default.
- W4312533318 hasConceptScore W4312533318C111919701 @default.
- W4312533318 hasConceptScore W4312533318C11413529 @default.
- W4312533318 hasConceptScore W4312533318C118505674 @default.
- W4312533318 hasConceptScore W4312533318C118615104 @default.
- W4312533318 hasConceptScore W4312533318C120174047 @default.
- W4312533318 hasConceptScore W4312533318C121332964 @default.
- W4312533318 hasConceptScore W4312533318C129782007 @default.
- W4312533318 hasConceptScore W4312533318C134306372 @default.
- W4312533318 hasConceptScore W4312533318C154945302 @default.
- W4312533318 hasConceptScore W4312533318C162324750 @default.
- W4312533318 hasConceptScore W4312533318C165801399 @default.
- W4312533318 hasConceptScore W4312533318C176217482 @default.
- W4312533318 hasConceptScore W4312533318C184898388 @default.
- W4312533318 hasConceptScore W4312533318C198043062 @default.
- W4312533318 hasConceptScore W4312533318C21547014 @default.
- W4312533318 hasConceptScore W4312533318C2524010 @default.
- W4312533318 hasConceptScore W4312533318C33923547 @default.
- W4312533318 hasConceptScore W4312533318C41008148 @default.