Machine Learning is a scientific discipline focused on the design and development of algorithms that improve their efficiency through experience (using either labeled or unlabeled observations). It draws on techniques and theoretical knowledge from the fields of artificial intelligence, logic, and statistics, forming a highly complex domain. Machine Learning is now a mature field of computer science, with robust theoretical models and results, as well as a wide range of applications in both industry and numerous research disciplines. It has also become very popular with the general public, thanks to the success of deep learning in game algorithms (Chess, Go, etc.) and the emergence of Data Science for the intelligent processing and analysis of massive datasets.
The A3 team is structured around three main research areas :
The central theme of the ADA research axis is representation learning from data and learners, for collaboration and transfer. It focuses on three multi-model learning paradigms based on three mathematical formalisms: collaborative/federated learning, transfer learning, and learning from multi-modal data through emerging formalisms (for learning) such as quantum formalism, optimal transport, and algebraic topology. The contributions of the ADA axis address both fundamental research and more applied studies, most often supported by academic and industrial collaborative projects in various fields such as healthcare, digital marketing and recommendation, complex system diagnostics, data quality and anonymization, social network analysis, and more.
Learning models expressed as explicit and explainable logical programs from graph-structured data. This axis specifically addresses topics such as relational learning and uncertain examples, probabilistic relational learning within a POMDP framework, graph abstraction, and abstract closed pattern mining.
Structure learning is a family of AI models that build and enhance numerous decision-making processes, offering significant potential for industrial applications and technology transfer. This axis addresses the simultaneous learning of data features and structures, multi/co-clustering, and structure learning using generative and/or probabilistic deep learning models. This family of models makes it possible to abstract large quantities of data—which would be unexploitable individually by human operators—by organizing them based on their homogeneities and differences. This axis focuses on multivariate time-series data, multimodal data processing, and large-scale learning.
Link to the 2017–2022 Activity Report and 2023–2028 Project form the A3 team.