Inria/CWI Associate team MDG-TAO

Goal

Our goal is to tackle Space Weather modeling by a synergetic combination of state-of-the-art simulations and Machine learning techniques. The huge amount of space missions data has not yet been systematically exploited in the current computational methods for Space Weather. Thus, we believe that this work will result in cutting-edge results and will open further research topics in Space Weather and Computational Plasma Physics.

Ongoing Projects

Feature learning from Solar images

Solar events automatic classification based on solar images

Dynamic forecasting of solar events

Solar wind and DST index prediction

Results

4 different auto-encoder architectures to automatically extract features for solar magnetograms

a baseline predictor of solar wind based on extracted features

a baseline classifier of solar events based on extracted features

A dynamic time lag forecasting system of the solar wind based on solar images

Events

Machine Learning in Heliophysics, 16-20 September 2019, Amsterdam

Lorentz Center workshop Space Weather: a Multi-Disciplinary Approach 25-29 September 2017

Publications

M. Chandorkar, C. Furtlehner, B. Poduval, E. Camporeale, M. Sebag, Dynamic Time Lag Regression: predicting What & When , ICLR (2020)

M. Chandorkar Machine Learning in Space Weather , Phd thesis, University of Eindhoven (2019)

M. Hajaiej(2017): Deep Solar Imaging for Geomagnetic Storms Prediction , master thesis report Ecole polytechnique (2017)

Chandorkar, Camporeale, Furtlehner and Sebag (2017): Bayesian Inference of Plasma Diffusion Parameters: An LSSVM based approach submitted to NIPS workshops on machine learning and physics.

M. Chandorkar, E. Camporeale, S. Wing (2017): Probabilistic Forecasting of the Disturbance Storm Time Index: An Autoregressive Gaussian Process approach, Space Weather, 15, 1004

Decelle, Fissore and Furtlehner (2017): Spectral Dynamics of Learning Restricted Boltzmann Machines, to appear in E. Phys lett. B.