AI Paper Atlas

Research map / Learning theory and algorithms / Clustering

Communication-Optimal Distributed Clustering

Jiecao Chen, He Sun, David Woodruff, Qin Zhang

NeurIPS 2016 · Learning theory and algorithms › Clustering · approach: Clustering approximation algorithms · 19 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
sparsifier-based distributed spectral clustering (sites send local spectral sparsifiers) instead of sending all datagraph spectral clustering · Sculpture, Twomoons, Gaussnormalized cut: at most ~2% larger than centralizedNo clear effect
blackboard recursive sampling instead of point-to-point sparsifier merging, as the number of sites growsgraph spectral clustering communication · Twomoons with 90 sitesedges communicated: MsgPassing communicates 9x edges of BlackboardImproved
fewer blackboard iterations (chain length 5 instead of 18) (ablation)blackboard sparsifier · graph clustering datasetsnormalized cut: ncut values divergeHurt
sampling only 3n edges per site (c=3) with one site in message passing (ablation)MsgPassing · graph clustering datasetsnormalized cut: ncut divergesHurt
increasing sampled edges per site (c) or number of sites s in message passing (ablation)MsgPassing · graph clustering datasetsnormalized cut: slight decreaseImproved
more blackboard iterations and more sampled edges per iteration (ablation)Blackboard · graph clustering datasetsnormalized cut: reduces ncutImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy