GeNeura at European Project MUSES Final Review

GeNeura’s members have been working in the three-year long, FP7 European project MUSES, which faced its last review last week at the European Commision Beaulieu Quarter Buildings in Brussels.

UGR was one of the partners participating in this project. More concretely, GeNeura’s members have contributed by leading WP2 – MUSES framework definition and integration during the completion of tasks to define the MUSES System Architecture. In addition, GeNeura’s research has been applied to the project in WP5 – Self-adaptive event correlation, lead by a Spanish security company S2 Grupo. The main purpose of this WP was to develop a system which, on the one side, uses event correlation to detect Security Policy violations and, on the other side, performs an analysis of all the data in the system and creates new Security Policies or enhances the existing ones. Different types of classification, rule association, and clustering algorithms, as well as Data Mining techniques, have been applied with satisfactory results. These results were specially welcomed by the comission, ponting that such a system will be very helpful to enhance security. Also, MUSES is an Open Software project, and you can contribute at https://github.com/MusesProject

The results were presented by S2 Grupo and GeNeura together. The slides are now published on Slideshare:

It has been a pleasure for GeNeura to work in MUSES

CTjLUuFWcAAhTqw.jpg:large

KohonAnts Slides (ALIFE XI)

Hello again to everyone!

These are the slides of the presentation of KohonAnts algorithm in ALIFE XI conference. ;)

It is an hybrid Ant Colony and Self-organizing Map algorithm for clustering and pattern classification.

A bit late, but I had some troubles with slideshare…

In any case…….Enjoy it. ;) :D

Here you can see an example of the evolution of the ants in the grid for the IRIS dataset:

Ants movement in the toroidal grid

Ants movement in the toroidal grid

Green class is quite similar to the other two classes, so it is difficult to get a fine cluster with it.

Thanks to Dave Oranchak. ;)