RNTI

MODULAD
Clustering Multi-Vues en utilisant la Représentation des Coordonnées Barycentriques
In EGC 2024, vol. RNTI-E-40, pp.353-360
Abstract
In this article, we address the problem of multi-view clustering, where attributes are decomposed into groups that provide complementary information. We present a new multi-view clustering approach BCmvlearn that combines the barycentric coordinate (BC) representation used in previous clustering work with the KMeans-based multi-view clustering RMKMC, which allows automatic updating of view weights. This approach reduces complexity without sacrificing clustering quality. In addition, our algorithm does not depend on the vector form of the original data, making it applicable to multimodal clustering.