Sublinear Algorithms for Correlation Clustering

Explore a lecture on sublinear algorithms for correlation clustering presented by Slobodan Mitrovic from UC Davis. Delve into the correlation clustering problem, which aims to cluster graph vertices to minimize between-cluster positive and within-cluster negative edges. Learn about a recent result achieving a 3+eps approximation using O(Delta/eps) LCA probes, O(log 1/eps) MPC rounds, and O(1/eps) expected update time in fully dynamic settings under oblivious adversary. Discover how this work relates to the seminal research on computing maximal independent sets in the LCA model. Gain insights into the technical analysis behind these results, based on joint work with Mina Dalirrooyfard and Konstantin Makarychev.

Sublinear algorithms for correlation clustering