Research

Research

I was researcher in Theoretical Computer Science (80--95). I was interested in Tree Automata (see the TATA book) and their applications (a paper on Set Constraints). Then I moved to Machine Learning (95--).  I studied Teaching Models using Kolmogorov Complexity; Learning from Positive Examples; Conditional Random Fields for Trees; Grammatical Inference Learning Algorithms for Trees and Boosting Algorithms for Multi-Task Learning. More recently I studied graph-based learning (see Hypernode Graphs and their application to the skill rating problem for multiple player games).

Now I am member of the MAGNET research team. I am involved in Natural Language Processing. I am interested in the combination of Language Models and Knowledge based models. I am member of the Impress project. We developed Mangoes, a Python toolbox for constructing and evaluating Language Models. 

I created the first research group -- called GRAppA -- in Machine Learning at Lille University and was head of this group from 1996 until 2013. I was also head of an INRIA research group -- called MOSTRARE -- from 2003 until July 2011. 

Publications