Yapay Zeka · 8 dk okuma · 20 Nisan 2026
Chromatic Clustering Requires New Algorithms to Match Standard Performance
Adding color constraints to correlation clustering increases computational difficulty; a new coupled approach recovers optimal approximation bounds.
Assigning semantic colors to clustering edges creates irreducible computational cost; coupled algorithms recover standard performance.
- — Standard correlation clustering achieves 2.06-approximation; chromatic variant has strict 2.11 lower bound.
- — Cross-edge chromatic interference forces color-independent algorithms to pay mismatch penalty absent in uncolored clustering.
- — Global Integrality Gap Decomposition Theorem decomposes total gap into standard gap plus irreducible chromatic penalty.
- — Two-color clustering gap reaches 2.0967, separating performance from standard case at L=2 colors.
- — Color-Coupled Correlation Clustering (C4) adds global constraint and correlated rounding to recover 2.06 approximation.
- — Neutral edges (color-mismatched) behave like negative edges under C4, bypassing uncoupled LP lower bound.
- — Theory validated on extremal instances, multi-relational networks, and fairness benchmarks with empirical gap matching predictions.
Sık sorulanlar
- Colors introduce cross-edge chromatic interference: edges whose color does not match the cluster's assigned color create unavoidable mismatch cost. In standard clustering, all negative edges incur the same penalty. With colors, neutral edges (color-mismatched) add an irreducible extra penalty that color-independent rounding algorithms cannot avoid, increasing the approximation lower bound from 2.06 to at least 2.11.