Unleash Big Data Clustering
summary
Re-implements DBSCAN to scale from small datasets to 400 million records by parallelizing the algorithm into a graph-based workflow. Describes using a graph algorithm to avoid O(n^2) pairwise complexity and shows an end-to-end solution that completes density-based clustering of very large datasets in a few hours.
AI-written summary · about
related talks
34mHDBSCAN, Fast Density Based Clustering, the How and the Why
35mClustering: A Guide for the Perplexed
54mDistributed Graph DBMSs — The Future for Analytics on Big Data?
48mBig Graph: Big Data applied to giant, dynamic graphs
42mDatabase Clusters: Making Complex Things Simple
25mThe Art of Clustering: The Good, The Bad and The Beautiful
01h 20mClustering Data
27mRapid deduplication and fuzzy matching of large datasets using Splink