Unverified50% confidenceFactExact time
The silhouette coefficient is a metric for evaluating clustering quality, ranging from -1 to 1, with values closer to 1 indicating tighter intra-cluster compactness and greater inter-cluster separation.
1
Sources
50%
Confidence
Long-term
Relevance
11/17/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedA spectral basis is a set of orthogonal basis vectors obtained through eigendecomposition, where the magnitude of the eigenvalues reflects the information content or importance of the corresponding direction.71% similarUnverifiedThe annotation granularity for deformation detection can progress from image-level binary classification labels, bounding boxes, and pixel-level semantic segmentation to multi-dimensional annotations such as deformation type, severity, and 3D depth estimation.67% similarUnverifiedSpectral concentration measures the degree of variance concentration through eigenvalue analysis of the feature covariance matrix, requires no labels, relies only on the features themselves, and can serve as an unsupervised diagnostic tool.65% similarUnverifiedCompared with adding a uniform edge bias to the scores, head-level score modulation allows different heads to attend to different types of edge relationships or subgraph patterns, enhancing the model's ability to express heterogeneous structural information and being naturally compatible with the block-wise computation flow of standard attention kernels.65% similarUnverifiedThe traditional approach computes centrality once on a homogeneous graph, with all nodes compared on the same scale. However, in a typed knowledge graph, computing them mixed together directly would obscure structural differences within categories.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1178175API
curl https://kongchang.com/api/v1/knowledge/claims/1178175MCP
get_claim(id=1178175)