12 November 2024 in Pattern Recognition
3 September 2025 in IEEE Transactions on Fuzzy Systems
To partition samples into distinct clusters, Fuzzy C-Means (FCM) calculates the membership degrees of samples to cluster centers and provides soft labels, gaining significant attention in recent years. However, existing FCM methods encounter the following challenges. First, traditional FCM focuses on learning membership degrees, neglecting the data similarity structures. Second, graph-based FCM typically separates graph construction from clustering, overlooking the knowledge interaction between graphs and clustering, obtaining suboptimal performance. Third, exploring the similarity structures among all samples is computationally expensive for large-scale tasks. To solve these dilemmas, we propose a scalable fuzzy clustering with collaborative structure learning and preservation (CSLP), which simultaneously leverages both cluster information and similarity structures to learn an optimal membership degree representation. Specifically, a self-weighted manner is devised to measure the sample importance, thereby reducing the adverse impacts of outliers. Moreover, the graph is updated according to the data similarities in the membership degree representation, such that CSLP collaboratively learns the graph and membership degrees in a mutually reinforcing manner. Thus, the similarity structures are fully explored during clustering processes and preserved in the learned membership degrees, enhancing the discrimination of clustering labels. To further improve efficiency, an acceleration solution is developed to reduce the computational cost of CSLP by propagating membership degrees from potential centers to samples, making CSLP scalable for large-scale tasks. An iterative strategy is designed to solve the formulated objective function. Extensive experiments demonstrate that CSLP outperforms other fuzzy clustering methods in terms of both effectiveness and scalability.
21 March 2024 in IEEE Transactions on Emerging Topics in Computational Intelligence
Data fusion has become an important task in multi-view learning. Previous methods suffer from the insufficient data fusion due to the following issues: (i) Several methods ignore the correlation and distinction among views and directly concatenate the features from different views; (ii) They involve intractable parameters to balance different views, degenerating the applicability of models; (iii) A fixed label matrix is used to guide feature fusion, overlooking the distances between different classes (i.e., inter-class distance) or within the same class (i.e., intra-class compactness). To overcome these problems, a novel fusion model is proposed to discriminate different views and samples in an adaptive manner, so as to effectively reduce the adverse impacts of low-quality views and outliers. In contrast to existing methods, a flexible regression target is designed to take full advantage of the label information of data, such that both the inter-class distance and the intra-class compactness are preserved. Benefiting from this, a compact and discriminative representation of multiple views is learned to maintain the consistent and complementary information of diverse views. Extensive experiments validate the effectiveness and the superiority of our proposed model.