The Signal and Statistics activity focuses on signal and image processing and statistical modeling. Research activities are inspired from data processing challenges in various application fields such as health engineering, nondestructive testing of materials, acoustics, remote sensing, astrophysics, transportation, electrical and mechanical engineering.
Signal processing methods rely on a wide range of mathematical tools such as multivariate statistics, numerical optimization, random matrix theory, sparse representation, Bayesian inference and tensor decomposition. This expertise allows us to propose solutions to big and possibly heterogeneous data analysis, statistical learning, data mining, temporally and spatially correlated signal analysis, optimal design of experiments, and inverse problems. The group is also interested in Algorithm-Architecture Matching issues, at the interface between signal processing and High Performance Computing. This activity aims at fully exploiting the significant potential of parallel computing of signal processing algorithms on GPU and FPGA hardware accelerators.
RESEARCH TEAMS IN SIGNAL PROCESSING AND STATISTICS
The Inverse Problem Team is positioned at the interface between physics and statistical signal and image processing. The research work relies on variational and Bayesian approaches for signal and image reconstruction. The main application fields include audio, nondestructive testing, and astrophysics.
The Modelisation and Estimation team aims to address massive data analysis problems based on mathematical tools such as multivariate statistics, robust statistics, and tensor analysis. The group is also interested in statistical approaches for optimal design of numerical experiments. The application fields include health, energy production, remote sensing, finance and statistical physics.