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...

Descrizione completa

Dettagli Bibliografici
Autori principali: Rebuffi, SA, Ehrhardt, S, Han, K, Vedaldi, A, Zisserman, A
Natura: Conference item
Lingua:English
Pubblicazione: IEEE 2021

Documenti analoghi