Progressive Cluster Purification for Unsupervised Feature Learning
This work addresses a specific bottleneck in unsupervised feature learning for representation models, offering an incremental improvement to clustering-based methods.
The paper tackles the problem of class inconsistent samples in clustering-based unsupervised feature learning by proposing Progressive Cluster Purification (PCP), which iteratively excludes noise samples during progressive clustering, resulting in significant improvements over baseline methods on common benchmarks.
In unsupervised feature learning, sample specificity based methods ignore the inter-class information, which deteriorates the discriminative capability of representation models. Clustering based methods are error-prone to explore the complete class boundary information due to the inevitable class inconsistent samples in each cluster. In this work, we propose a novel clustering based method, which, by iteratively excluding class inconsistent samples during progressive cluster formation, alleviates the impact of noise samples in a simple-yet-effective manner. Our approach, referred to as Progressive Cluster Purification (PCP), implements progressive clustering by gradually reducing the number of clusters during training, while the sizes of clusters continuously expand consistently with the growth of model representation capability. With a well-designed cluster purification mechanism, it further purifies clusters by filtering noise samples which facilitate the subsequent feature learning by utilizing the refined clusters as pseudo-labels. Experiments on commonly used benchmarks demonstrate that the proposed PCP improves baseline method with significant margins. Our code will be available at https://github.com/zhangyifei0115/PCP.