Professor Ludovic Denoyer




Ludovic Denoyer is a Research Scientist at Criteo and a Full Professor in Machine Learning at Sorbonne’s University (France)

He obtained his Master of Artificial Intelligence in 2001 and his Ph.D. thesis in 2004 on the problem of classification with structured inputs. He then focused on two main research topics:

- representation learning for structures, and particularly complex graphs (multi-relational, dynamic,
heterogeneous) and social networks

- budgeted reinforcement learning models with applications to text classification, image classification, features acquisition and structured output prediction.

His research is now mainly centered on the desire to develop human-machine interaction systems. Since October 2017, he is also a part-time researcher at Criteo on the topic of statistical machine learning from human traces. During his career, he was the organizer of many international challenges financed by European networks of excellence (PASCAL and DELOS), and also created reference datasets (Wikipedia XML Corpus in 2003 and WebSpam dataset in collaboration with Yahoo! in 2007/2008) in the Information Retrieval and Machine Learning domains. He has supervised dozens of Ph.D. students, the current ones being involved on different topics -- hierarchical reinforcement learning, adversarial models for disentanglement, spatio-temporal neural networks, style extraction in reinforcement learning, ... He is also participant/head of many national and international research projects. In addition, Ludovic Denoyer is also co-creator and co-head of the Data Science Master (about 100 students each year) and of the new lifelong training curriculum (2018). In the meantime, he has been ‘chargé de mission’ Open Research reporting to the president of the university (2017), and member of the ‘France Artificial Intelligence’ thinking group.