oai:HAL:hal-04105310v3
HAL CCSD
technologies: computer sciences
2023
12/15/2023
International audience; We introduce a differentiable clustering method based on stochastic perturbations of minimum-weight spanning forests.
This allows us to include clustering in end-to-end trainable pipelines, with efficient gradients.
We show that our method performs well even in difficult settings, such as data sets with high noise and challenging geometries.
We also formulate an ad hoc loss to efficiently learn from partial clustering data using this operation.
We demonstrate its performance on several data sets for supervised and semi-supervised tasks.
Stewart, Lawrence,Bach, Francis, S,López, Felipe Llinares,Berthet, Quentin, 2023, Differentiable Clustering with Perturbed Spanning Forests, HAL CCSD