SequeL is a research group working in the field of machine learning; more specifically, SequeL is dedicated to the study of the problem of sequential decision making under uncertainty, that is, the study of how an “agent” having a goal to fullfill can learn an optimal behavior to achieve this goal in an unknown environment. SequeL is composed of two dozens members. Activities range from foundations of learning to algorithm design, and transfer towards companies. Questions are studied such as “What can a Turing machine learn efficiently? and in which conditions?”. Or, in a budget context, “Given an amount of computational resource, how close to the optimal behavior can an algorithm reach?”, finally application oriented questions such as those related to computational advertizing and recommendation systems for e-commerce websites, are also studied.

SequeL has led to the multi-awarded Crazy Stone go playing program. Some SequeL PhD students have been awarded the Gilles Kahn award, the Jacques Neveu award and the ECCAI award. We won the ICML 2011 Exploration vs. Exploitation challenge, and the ACM RecSYS 2014 challenge (both challenges on recommendation systems). SequeL expertize has led to collaborations with international companies like Orange Labs, Intel, Technicolor, Deezer and also with national and local SMEs.