MiCE: Mixture of Contrastive Experts for Unsupervised Image Clustering
This work addresses the problem of unsupervised image clustering for computer vision researchers, offering an incremental improvement by integrating existing techniques.
The authors tackled unsupervised image clustering by proposing MiCE, a framework that combines contrastive learning with a latent mixture model, achieving significantly better results than previous methods on four natural image datasets.
We present Mixture of Contrastive Experts (MiCE), a unified probabilistic clustering framework that simultaneously exploits the discriminative representations learned by contrastive learning and the semantic structures captured by a latent mixture model. Motivated by the mixture of experts, MiCE employs a gating function to partition an unlabeled dataset into subsets according to the latent semantics and multiple experts to discriminate distinct subsets of instances assigned to them in a contrastive learning manner. To solve the nontrivial inference and learning problems caused by the latent variables, we further develop a scalable variant of the Expectation-Maximization (EM) algorithm for MiCE and provide proof of the convergence. Empirically, we evaluate the clustering performance of MiCE on four widely adopted natural image datasets. MiCE achieves significantly better results than various previous methods and a strong contrastive learning baseline.