RESEARCH

 

Home  -  Curriculum Vitae  -  Publications  -  Research

Last update: September 2014

 

 

Michaël Aupetit

 

Qatar Computing Research Institute (http://qcri.org.qa/)   

Computational Science and Engineering

10th Floor, Tornado Tower,

PO Box 5825, Doha, Qatar

http://michael.aupetit.free.fr/

(+974) 445 47150 –  michael.aupetit@qf.org.qa

 

Research

 

My research focuses on designing new techniques to take up challenging scientific issues at the crossing of Computational Topology, Visual Analytics and Machine Learning, to support humans in their decision facing (big) data coming from the monitoring of (complex) systems.

I design intelligent user-centric decision support systems to make easier Human Machine Interaction. I open the black-box of Machine Learning techniques to allow human users to get access to the knowledge acquired by the machine, in order to get the best combination of machine’s speed and accuracy and human intelligence. I designed techniques in the following domains:

 

Visual Analytics based on Dimensionality Reduction

This research deals with making interpretable and usable the scatter plot projections of multivariate data for exploratory data analysis and interactive clustering.

Main publications: [InfoVis2012] [CGF2011] [EuroVAST2010] [Neurocomputing2007]

 

 

Description: figure3

Data Mining and Machine Learning

This research deals with exploratory data analysis, clustering and classification techniques relying on automatic machine learning, computational geometry and topology.

Main publications: [Neurocomputing2009-2008-2007-2005] [NeuralNetworks2007-2002] [NIPS2006]

 

 

Distributed computing and multi-agent systems

In this research I studied how distributed computing can be used to map a building and guide people and robots within it.

Main publications: [Patents to be submitted 2014]

 

 

 

Interpretability of Fuzzy Rule Based Systems

In this research I study Fuzzy Rule Based Systems able to preserve knowledge interpretability while optimizing their parameters, and FRBS within a probabilistic framework to use standard Machine Learning techniques for their optimization.

Main publications: [LFA2013] [LFA2006][IPMU2014] [Patent pending 2012]

 

 

 

 

 Description: france_ml3