Nistor Grozavu Personal Website

                                                              LIPN  A3  


During my thesis I worked within the Infom@gic Project.

In this project I worked on the 'Learning/Clustering of multimodal datasets and Clustering Fusion', and we (LIPN & THALES) proposed the next Content-based image retrieval System.

Here you will find a video which shows our proposed Content-based image retrieval System:

 This system uses the Weighted Topological Learning on a multimedia library (wikipedia images).  It is based on my proposed topological methods and uses a clustering fusion technique (Relational Analysis) which allows to choose the best map. It allows a quick browsing of a images dataset and images searching.

This approach uses the Weighted Self-Organizing principle and builds a 3D (hierarchical) view of images.

For further informations about this System contact me at Nistor.Grozavu [@]


One of my recent works was the clustering analysis using topological learning/clustering and variables selection/detecto. Here, an example of the clustering charachterization/analysis on the waveform dataset (varaibles 1-19 -relevant variables; 20-40 are noisy variables):


Another interest topic is the topological clustering collaboration or distributed clustering: