MIA-Paris (Applied mathematics and computing-Paris)February 11 2022
UMR 518, AgroParisTech, INRAE, Université Paris-Saclay
General scientific orientation
The Mia-Paris unit brings together statisticians and computer scientists specialising in modelling and statistical and machine learning for biology, ecology, environment, agronomy and agri-food. Their skills include statistical inference methods (stochastic modelling, latent variable models, Bayesian inference, statistical learning, model selection...), and algorithmic methods (generalization, domain transfer, knowledge representation). The lab develops original statistical and computational methods that are generic or motivated by specific problems in life science. Its activities are based on a good culture in the target disciplines : agronomy, agri-food, molecular biology, systems biology, ecology, environment.
The lab is structured into three research teams :
- Study climatic, ecological, environmental risks and develop statistical methods to address these areas where data and their structures are increasingly complex.
- Research topics : spatial and spatio-temporal statistics (hierarchical Bayesian models, point processes, conditional process simulations), multivariate and spatialized extremes, numerical experiments, uncertainty propagation and Bayesian decision theory, analysis and inference of random graphs, trajectory modeling.
- Develop and disseminate statistical modelling and machine learning methods for the analysis of (meta)genomic, genetic or metabolomic data.
- Research topics : network analysis and inference, modelling and statistics of large-scale data, segmentation and time series, supervised (classification, regression) and unsupervised (dimension reduction, clustering) learning.
- Enable the exploitation of data from multiple and heterogeneous sources, even in flow, based on the enlightened choice of shared and multi-scale semantic representations, in order to contribute to the enrichment of expert knowledge in the field of food and life sciences.
- Research topics : modeling and analysis of heterogeneous multisource data, human and machine multi-expertise (taking into account semantics), collaborative and incremental learning methods, combinatorial optimization, representation and integration of data and knowledge on the Semantic Web and the Web of linked data, learning of probabilistic models, study of causality.