Online learning & ensemble forecasts | Renewable energies
Improved forecasts and probabilistic forecasts based on multiple deterministic forecasts, using online learning (sequential aggregation). Application to weather forecast for renewable energies.

Meta-modeling & model reduction | Air quality
Design of surrogate models for complex urban pollution models. Speed-up factor of 10,000 or more, using dimension reduction and Gaussian processes or RBFs.

Assimilation of mobile data | Urban noise
Assimilation of noise measurements collected by smartphones in order to improve urban noise maps and provide low noise exposure routing to smartphones’ users.

Danger mapping using deep learning | Wildfires
Deep learning to replace a complex fire propagation model, with scalar and image inputs. Computation of danger maps for use by fireworkers.