Bingbing Jiang

Discovering complex-shaped clusters while maintaining computational efficiency remains a fundamental challenge in clustering. Herein, we propose RepPHC, a parallel hierarchical clustering method based on dynamically updated representativeness-aware inter-cluster affinities. RepPHC constructs a sparse $k$NN graph, computes local point affinities from neighborhood proximity and density similarity, and aggregates them across clusters using sample representativeness. As clusters evolve, sample representativeness is dynamically updated, enabling inter-cluster affinities to adapt to changing cluster structures and providing more reliable merging decisions. At each hierarchy level, multiple strong and non-conflicting cluster pairs are merged simultaneously, substantially reducing the number of agglomeration rounds. Finally, we design a cluster-persistence criterion to select stable clusters from the hierarchy as the final clustering result and identify their representative centers. Benefiting from sparse graph operations, recursive affinity updates, and parallel merging, RepPHC achieves an overall time complexity of O(TkNN+nklog(n)). Extensive experiments on synthetic and real-world datasets demonstrate that RepPHC effectively recovers complex-shaped clusters, reliably identifies cluster centers, and accurately detects stable clusters.

Multi-view fuzzy clustering has garnered growing interest for its ability to effectively model soft membership relationships among samples. However, most existing methods typically compute sample membership to cluster centers directly, lacking a learnable intermediate representation layer to refine the modeling and propagation of membership information. Furthermore, some visible-hidden collaborative clustering frameworks rely on linear matrix factorization to extract a hidden view, which fails to accurately capture the nonlinear structure of the data. Therefore, this paper proposes a visible-hidden collaborative multi-view fuzzy clustering via adaptive quadratic matrix factorization and membership propagation (VCMFC-AM). This method first extracts the hidden view through an adaptive quadratic matrix factorization mechanism, which explicitly models nonlinear data structures by constructing quadratic terms and introduces adaptive weights based on reconstruction error to fuse multi-view representations. Building on this foundation, this paper further designs an end-to-end collaborative fuzzy clustering framework that introduces learnable anchors as an intermediate representation layer, establishing a two-layer membership propagation path involving sample-anchor-cluster. Additionally, the framework integrates graph regularization constraints guided by both visible and hidden views to enhance the discriminative power of the clustering results. Extensive experimental results demonstrate that VCMFC-AM significantly outperforms existing state-of-the-art methods in terms of clustering performance. Furthermore, survival curve analysis on cancer multi-omics datasets validates the method’s potential application value in clinical prognosis prediction tasks.

Clustering on high-dimensional and low sample size (HDLSS) data remains a critical, persistent challenge where extreme sparsity and noise confound cluster analysis. This creates a dilemma: spectral methods fail as distance metrics degrade, while deep clustering tends to over-fit scarce data. To break this dilemma, a Self-Enhanced Density Clustering (SEDC) framework that integrates the cluster structure discovery and embedding representation learning into an iterative enhancement process is proposed in this paper. Specifically, SEDC uses adaptive density-derived centroids to parameterize probabilistic soft labels, which in turn supervise a lightweight multilayer perceptron (MLP) to learn the low-dimensional embedding from data. The resulting embedding provides a refined metric space for further generating superior labels in the subsequent interaction process. This feedback forms a mutual reinforcement that progressively enhances the discrimination of embedding while rigorously mitigating over-fitting. Extensive experiments on 43 challenging HDLSS datasets demonstrate state-of-the-art performance, substantially outperforming popular clustering methods. This work delivers a principled and promising solution for robust data clustering in HDLSS situations.

With the advancement of brain–computer interfaces (BCI), motor imagery (MI) electroencephalogram (EEG) decoding can greatly benefit from spatial filtering features derived from common spatial patterns (CSP). However, CSP‐based features often exhibit high redundancy and intersubject variability. These limitations make the feature selection methods based on sparse learning difficult to effectively balance the heterogeneous contributions of different temporal and spatial components. Moreover, these models tend to prioritise features with larger coefficients, potentially overlooking intrinsic feature importance and compromising the quality of the selected feature subset. To address these issues, we propose an Adaptive Sparse Group Lasso (ASGL) method for structured feature selection, designed to enhance discriminative CSP features whilst suppressing irrelevant components. The proposed method partitions EEG signals into consecutive segments using a sliding window, treating each as a separate feature group. Benefiting from this, the importance of features at both the group level and the within‐group level can be effectively quantified through mutual information and copula mutual information, thereby assigning adaptive weights for selective penalisation within the model. This weight construction strategy preserves important features from relevant time intervals and frequency bands. The resulting optimization problem is solved efficiently via the alternating direction method of multipliers (ADMM). Evaluations on simulated and real‐world datasets demonstrate that the proposed ASGL outperforms existing methods.