I did my PhD on the learning of context-free languages, from positive or positive and negative non-structured sequences (strings). This work is known as the starting point of what is now called Distributional Learning(Alexander Clark has a nice and short presentation of the main ideas). When I arrived in Marseilles in 2007, I continued this line of work (moving to graph grammars and to string-to-string function learning, for instance), but my interest grew also into more common machine learning approaches. The proximity of a bunch of brilliant and friendly researchers interested in statistical machine learning, natural languages processing, or multi-media indexing and retrieval, created an exiting environment.
My latest work deals with the interpretability and explainibility of deep learning black-boxes, in particular of Recurrent Neural Networks.
One can find more detail in my personal résumé [Last update: Sep 2018].
- With my colleague Stéphane Ayache and our MSc student Noé Goudian, we have a paper accepted at ICGI’18! It deals with Black Box (RNN) interpretability.
- I am involved in the new Master program Master Informatique, parcours Intelligence Artificielle & Apprentissage Automatique (IAAA) (Artificial Intelligence & Machine Learning) that opens in September 2018.
- I am the chair of the International Conference on Grammatical Inference steering committee since 2014.
- I organized with Jeffrey Heinz, Guillaume Rabusseau, and Matteo Sammartino the second edition of the LearnAut, workshop at FLoC 2018. It is on learning, logic, automaton, and other related topics.
- With my colleague Cécile Capponi, we gave a week long professional formation at Leroy Merlin headquarter, on “Introduction to Machine Learning: from bases to deep learning”
- I participated to the organisation of Quatrième cycle de conférences MPCI on “L’informatique au coeur des sciences” (Computer Science in the Core of Sciences). It’s in Marseille on 22-24 May 2018
- I recently participated to the creation of a toolbox for the spectral learning of weighted automata (scikit-splearn): everything is described here.
- I am part of the organisation of the Treize Minutes Marseille. The 2017 edition was held on November 30th and the 2018 will be announced soon, stay tuned! The videos of previous editions are available on the website.
- I organized with Borja Balle and Leonor Becerra the first edition of the LearnAut workshop at LICS 2017 (it was in Reykjavik, Iceland, on June 19th 2017)
- I co-organized the Sequence Prediction Challenge (SPiCe) a machine learning competition about sequential predicition. For the competition I developped a spectral learning toolbox with François Denis, Denis Arrivault, and Dominique Benielli called Sp2Learn.
- I was a member of the NIPS 2016 reviewer team.
- I was a member of the program committee of the International Conference in Grammatical Inference (ICGI 2016).
- I was a member of the program committee of the francophone conference in Machine Learning CAp 2016 and in charge of part of its organisation. It was nice to see you all during July 2016 in Marseille!
- I gave a talk (’causerie’) for the “fête de la science” at the village of science on October 10th 2015. It took place at villa Méditerranée.
- I had a sabbatical in 2012 (thanks CNRS!) and passed the year at the University of Maryland, Baltimore County as a visiting researcher.
- The web site of the Master I2A that I was in charge of is available here.
- The PASCAL2 Bootcamp is over but you can find photos and more on the facebook page.
- I was a member of the program committee of the 12th International Conference on Grammatical Inference.
- I was a member of the organisation and scientific committees of the PAutomaC competition.