LSD-C: linearly separable deep clusters
We present LSD-C, a novel method to identify clusters in an unlabeled dataset. Our algorithm first establishes pairwise connections in the feature space between the samples of the minibatch based on a similarity metric. Then it regroups in clusters the connected samples and enforces a linear separat...
Autori principali: | Rebuffi, SA, Ehrhardt, S, Han, K, Vedaldi, A, Zisserman, A |
---|---|
Natura: | Conference item |
Lingua: | English |
Pubblicazione: |
IEEE
2021
|
Documenti analoghi
-
Automatically discovering and learning new visual categories with ranking statistics
di: Han, K, et al.
Pubblicazione: (2020) -
Semi-supervised learning with scarce annotations
di: Rebuffi, SA, et al.
Pubblicazione: (2020) -
AutoNovel: automatically discovering and learning novel visual categories
di: Han, K, et al.
Pubblicazione: (2021) -
Learning to discover novel visual categories via deep transfer clustering
di: Han, K, et al.
Pubblicazione: (2020) -
Efficient parametrization of multi-domain deep neural networks
di: Rebuffi, S, et al.
Pubblicazione: (2018)