Research themes

I'm interested in machine learning and data mining. I've a mixed background in mathematics and computer science, and I tend to be more interested by the statistical learning point of view than by a more database and mining way of approaching problems. The majority of my recent works concerns exploratory problems, from information visualization to clustering. I like also supervised problems in which case I'm interested in feature selection and in learning theory.

Most of my publications are available on line, together with software and data sets. Some of my talks are also available.

Latest research news

Research highlights

Recent events

I was a member of the jury of the PhD thesis defense of Bertrand Lebichot.
I gave a seminar on my recent paper on temporal graph at the Machine Learning Group of UCL.
My colleague Joseph Rynkiewicz defended is Habilitation à diriger des recherches. I was a member of the jury.
I gave an invited talk at the 3rd Big Data Conference of Linnaeus University
My colleague Pierre Latouche defended his Habilitation à diriger des recherches. I was one of the reviewer.
Marco Corneli defended his thesis
Our special issue of the Journal de la Société Française de Statistiques on Humanities and Statistics has been published.

PhD students



Editorial activities

From 2007 to 2015, I was associate editor of Neurocomputing and of Neural Processing Letters. Considering the increasingly aggresive behavior of commercial publishers and their profit maximization strategies, I've decided to resign from those positions at the end of 2015.

I'm a member of the scientific committee of ESANN.

I've reviewed literally hundreds of papers (around 80 journal papers, roughly 40 papers for ESANN alone, and dozens of papers for ICML, ECML, PKDD, NIPS, ICANN, IWANN, etc.). In 2012, I reviewed more than 45 papers which made me wonder whether I might be reading more papers for reviewing purpose than for keeping in touch with recent results. This is a bit frightening. 2013 was not really better with 52 papers (I'm cheating a bit by counting 7 revised papers, but still…).