Centered Kernel Alignment (CKA)

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• Centered Kernel Alignmentmethod0.881
  Standard alignment metric cited and compared against; measures global kernel similarity between representations
• Centered Kernel Nearest-Neighbor Alignmentmethod0.835
  Modified CKA metric that restricts cross-covariance to nearest neighbors; introduced in this paper's appendix
• CKA shows a very weak trend of alignment between models even within modality, compared to mutual k-NN which shows stronger trendsfinding0.749
  Explains why mutual k-NN was chosen over CKA as primary metric
• Data-Centric Alignmentconcept0.725
  Alignment approach that focuses on curating or modifying training data; the paper bridges this with interpretability methods.
• Centerednessconcept0.724
  The defining mark of a center: the appearance of being a focal zone within a larger whole.
• As number of nearest neighbors k decreases in CKNNA metric, cross-modal alignment trend becomes more pronounced across both models and tasksfinding0.708
  Shows cross-modal alignment is primarily local rather than global
• Centers are above all, labile, they are foci of wholeness, they are not things, but regions, qualities, focal points of centeredness.quote0.699
  Key definition capturing the non-atomic, relational nature of centers as fields rather than objects.
• centered (living geometry)concept0.699
  The quality of a geometry that has life, characterized by strong centers; the talisman for recognizing living structure.