Answers are grounded in this paper and cite their sources.
Anchor-graph spectral clustering is widely adopted for large-scale clustering, yet existing anchor-based schemes suffer from two practical limitations: clustering quality is highly sensitive to anchor representativeness, and the anchor count must be manually predefined. Drawing on granular ball (GB) computing, we propose GB-ASC, an efficient spectral clustering method based on adaptive granular ball anchors. Our method automatically determines both the number and distribution of anchors from the intrinsic data structure, faithfully preserving complex data topology. Specifically, compact granular balls are adaptively generated via intra-ball distance minimization; their centers serve as anchors to construct a sparse anchor graph, and singular value decomposition (SVD) derives spectral embeddings for clustering. Extensive experiments on synthetic and real-world datasets demonstrate that GB-ASC outperforms state-of-the-art competitors in accuracy and robustness, validating its effectiveness on large-scale complex data.
No takes yet. Share an insight, caveat, or question.
Liu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: