Qarma's home

Research themes

statistical learning: domain adaptation, ranking, concentration inequalities, generalization bounds, metric learning
grammatical inference: stochastic rational languages (on strings and trees), weighted automata, learnable grammars
multimedia mining: fusion in multimodal learning, semi-supervised learning, multilabel classification
signal processing (with the signal and image processing group of LATP): we work on drawing connections between machine learning and signal processing (algorithms, theory, etc)

2010-2011, Grabbr project: "GRAph-Based Brain Reading." Interdisciplinary CNRS program "Neuroinformatics and computational neurosciences".
2010-2011, BISOU project (Gdr ISIS): "BiStochastic Optimization and mUltiple kernel learning".
Up until 2013: Member of PASCAL 2 Network of Excellence.
2009-2012, ANR ASAP: Statistical Learning with Deep Architectures.
2009-2012, ANR DECODA: "Dépouillement de conversations de centres d’appels".
2009-2012, ANR Edylex: "Enrichissement Dynamique de Lexiques".
2009-2013, ANR LAMPADA: "Learning Algorithms, Models an sPArse representations for structured DAta".
2010-2013, ANR PERCOL: "Reconnaissance de personnes dans des contenus audiovisuels".
2009-2012, ANR SEQUOIA: "Analyse syntaxique probabiliste à large couverture du français".
2010-2013, ANR Videosense: "Reconnaissance multimodale de concepts enrichis (statiques, dynamiques, émotionnels) dans des vidéos multilingue au travers de langages pivots".