Authors: Jing Pan, Yuhua Qian, Feijiang Li, Qian Guo

Abstract:

Reasonable feature representation plays an important role in improving the performance of clustering algorithms. However, recent deep clustering studies only focusing on feature representation at the pixel level leads to feature representation with low discrimination. Our key insight is that considering local-topology information between images would help to get a highly discriminative representation, and therefore we design a replacement strategy to find local-topology representation of data, and propose a two-stage image deep clustering algorithm based on local-topology embedding called ITEC. Specifically, we take advantage of data augmentation technique to improve the generalization performance of the learning models; then local-topology representation of data is embedded into the representation of data itself, so as to better complete tasks of image clustering. Extensive experiments demonstrate that local-topology information effectively promotes the performance of deep clustering significantly.

Keywords： Deep clustering; Autoencoder; Local-topology embedding; Data augmentation; Unsupervised learning

Image deep clustering based on local-topology embedding.pdf

查看详情

Fri Dec 24 15:35:00 CST 2021