Graph based Consistency Learning for Contrastive Multi-View Clustering
Abstract: Multi-View Clustering (MVC) aims to mine complementary information across different views to partition multi-view data more effectively and has attracted considerable interest. However, existing deep multi-view clustering methods frequently neglect the exploration of structural information within individual view and lack the learning of structural consistency among views, which results in limitations in the clustering performance. In this paper, we introduce a novel multi-view clustering framework based on graph consistency learning to address this issue. Specifically, we design intra-view graph contrastive learning to uncover structural information within each view and achieve structural conscistency objectives through cross-view graph consistency learning. Additionally, to address the conflict between different learning objectives when trained in the same space, we introduce two new feature spaces, one for cluster-levcel contrastive learning and the other for instance-level contrastive learning. Subsequently, to make the most of discriminative information from all views, we concatenate high-level features from all views to form global features and employ self-supervision to promote clustering consistency across different views. Experimental results on several challenging datasets demonstrate the outstanding performance of our proposed method.
Primary Subject Area: [Content] Multimodal Fusion
Relevance To Conference: Multimodal data typically consists of multiple views or sources, each representing different data modalities (such as images, text, audio, etc.). Multi-view clustering aims to leverage the complementary information across these different views to effectively partition the data. This work proposes a novel multi-view clustering method called GC-CMVC, aiming to enhance the effectiveness of partitioning multi-view data by leveraging complementary information across different views. Existing deep multi-view clustering methods often overlook the structural information within individual views and fail to ensure structural consistency among views, leading to limitations in clustering performance. To address this issue, GC-CMVC incorporates graph consistency learning. It introduces intra-view graph contrastive learning to reveal structural information within each view and achieve structural consistency objectives through cross-view graph consistency learning. Additionally, to mitigate conflicts between different learning objectives within the same space, two new feature spaces are introduced: one for cluster-level contrastive learning and the other for instance-level contrastive learning. Furthermore, the method utilizes self-supervision to promote clustering consistency across different views by concatenating high-level features from all views to form global features. Experimental results on challenging datasets demonstrate the superior performance of GC-CMVC in multi-view clustering tasks.
Submission Number: 2887
Loading