Representation Learning for Clustering via Building Consensus
This work addresses the problem of enhancing clustering robustness and performance for image data, particularly in real-world scenarios with distribution shifts, though it is incremental by building on existing consistency notions.
The paper tackles unsupervised representation learning for image clustering by introducing consensus consistency, which ensures similar partitions across variations in representation space, clustering algorithms, or initializations, and integrates it with existing consistencies in an end-to-end framework. The result is improved clustering performance over state-of-the-art methods on four out of five image datasets.
In this paper, we focus on unsupervised representation learning for clustering of images. Recent advances in deep clustering and unsupervised representation learning are based on the idea that different views of an input image (generated through data augmentation techniques) must be close in the representation space (exemplar consistency), and/or similar images must have similar cluster assignments (population consistency). We define an additional notion of consistency, consensus consistency, which ensures that representations are learned to induce similar partitions for variations in the representation space, different clustering algorithms or different initializations of a single clustering algorithm. We define a clustering loss by executing variations in the representation space and seamlessly integrate all three consistencies (consensus, exemplar and population) into an end-to-end learning framework. The proposed algorithm, consensus clustering using unsupervised representation learning (ConCURL), improves upon the clustering performance of state-of-the-art methods on four out of five image datasets. Furthermore, we extend the evaluation procedure for clustering to reflect the challenges encountered in real-world clustering tasks, such as maintaining clustering performance in cases with distribution shifts. We also perform a detailed ablation study for a deeper understanding of the proposed algorithm. The code and the trained models are available at https://github.com/JayanthRR/ConCURL_NCE.