The k-NN Identity Theorem Local Coherence and Classification Accuracy
Abstract
We present the k-NN Identity Theorem: for k = 1, the semantic tolerance τ∗ equals the leave-one-out accuracy of 1-NN exactly. For k > 1, we derive concentration bounds under diagonal dominance conditions. We validate empirically on 6,743 datasets.
Fields of Study & Subjects
Machine Learning and Data Classification
Data Mining and Machine Learning Applications
Machine Learning and Algorithms
Publication details
- Journal / source
- Zenodo (CERN European Organization for Nuclear Research)
- Publication date
- 29 January 2026
- Document type
- Preprint