Scientific project for testing new large scale clustering/quantization algorithms. These algorithms are typically distributed and asynchronous. The code is written in C#/.NET (4.0) and runs on Microsoft Azure Platform. The project is built using Lokad.Cloud which provides an O/C (Object to Cloud) Mapper and an 'execution framework' that leverages low level technicalities of MS Azure.

This project follows theoretical work made by Patra that can be found here http://arxiv.org/PS_cache/arxiv/pdf/1012/1012.5150v1.pdf This project is also naturally linked to the first work of Durut and Rossi on Azure implementation of the popular batch KMeans algorithm http://apiacoa.org/publications/2010/azure-k-means.pdf

This work has been funded by Lokad SAS, http://www.lokad.com/