Robust Deep Density Models for High-Energy Physics and Solar Astronomy
ROLE OF THE INSTITUTE FOR DATA SCIENCE
> Predictive and generative models for solar flare forecasting
> Models for solar active regions
> Algorithms for searching unknown solar features
Project lead at I4DS: André Csillaghy
Funding: Swiss National Science Foundation SNSF (SINERGIA)
Duration: 3 years
Keywords: big data, data science, machine learning, deep learning, heliophysics, high energy physics
During the last decade, the amount of data available to scientists has increased enormously. New infrastructures, such as the Large Hadron Collider (LHC), and a new generation of solar observatories, such as the Solar Dynamics Observatory (SDO), produce data on a scale that, with existing methods, cannot be exploited to their full extent. Finding the rare event of interest is like trying to find a needle in a hay stack.
This project is about new methods in machine learning resulting in better behaviour when dealing with rare events. The objectives are to create better forecasting tools, generative models and anomaly detectors to be applied in the fields of High Energy Physics and heliophysics.
PEOPLE @I4DS WORKING ON RODEM
OPEN RESOURCES AND RESULTS
ATLAS event display and visualization of the Standard Model (SM) of High Energy Physics and hypothetical extensions Beyond the Standard Model (BSM), including potential machine learning solutions. Image: Team AstroSignals
The ability to predict solar flares is vital for taking mitigation measures on Earth in time. At a more fundamental level, flare prediction can lead to a deeper understandig of the dynamics of the Sun. Above, a solar flare in ultraviolet wavelength by the space telescope Solar Dynamics Observatory SDO, August 9, 2011. Image: NASA/SDO
Spectra of a solar flare by the space telescope IRIS. Image: Brandon Panos, I4DS