descriptionHeuristic for k-medoids classification using MPI
last changeFri, 22 Jul 2016 16:26:26 +0000 (18:26 +0200)

Parallel clustering with a k-medoids algorithm

Joint work with Jairo Cugliari

This C program runs the k-medoid algorithm on several subsets of one (presumably big) dataset. The computed medoids are then merged iteratively, until we get a final set of k centers.

The folder “communication/” contains latex sources (and generated pdf files) of a short paper submitted to the Journées de Statistique in Rennes, France (2014), and also the slides presented at this event.

The other folder contains all the C code; but not the EDF (french electricity company) datasets, because they are not public. Since the (de)serialization process in code/src/TimeSeries/ is tailored for these data, it is necessary to adapt this small part of the code to use any other custom time-series files.

2016-07-22 Benjamin Auderfix accent (TODO: fix gitweb...) master
2016-07-22 Benjamin Auderreadme.md
2015-09-29 Benjamin Auderupdate
2015-02-05 Benjamin Auderupdate README
2015-02-05 Benjamin AuderAdd README.html
2015-02-05 Benjamin Auderrename latex --> communication
2015-02-04 Benjamin Auderrecompile slides (JdS 2014)
2015-02-04 Benjamin Auderadd code/_ReadMe. Clean latex/ folder
2015-02-04 Benjamin Audercommit last state
4 years ago master