Presentation Information
[PEM10-P05]Development of Classification and Regression Models for SEP Events Forecasting with Machine/Deep Learning Methods
*Yuta Kato1,2, Kanya Kusano2, Naho Fujita1, Kentaro Doi1, Akira Tsuda1, Hiroyuki Kawasaki1, Tomoki Moriyama2, Tomoki Kubo3, Chihiro Mitsuda1,2 (1.Fujitsu Limited, 2.Institute for Space–Earth Environmental Research, Nagoya University, 3.Graduate School of Science and Technology, Electrical and Information Engineering, Niigata University)
Keywords:
Solar Energetic Particles,Explainable AI,Machine Learning,Deep Learning,Timeseries Forecasting
Fujitsu and the Institute for Space–Earth Environmental Research (ISEE), Nagoya University, have been conducting joint research since September 2023 to establish space weather forecasts for the safety of human activities in space and on the Moon.
Solar Energetic Particle (SEP) events, which occur in conjunction with solar flares and coronal mass ejections, are significant space weather phenomena known to impact human health, spacecraft, and ground-based systems.
This presentation includes the current results and ongoing status as follows.
(1) SEP event occurrence prediction using a classification task with explainable AI (WideLeaning)
(2) SEP event particle flux, fluence, and time-series forecasting using a regression task with the machine and deep learning methods
We discuss possible SEP event prediction schemes with these developed models and the future prospects of this industry-academia collaboration.
Solar Energetic Particle (SEP) events, which occur in conjunction with solar flares and coronal mass ejections, are significant space weather phenomena known to impact human health, spacecraft, and ground-based systems.
This presentation includes the current results and ongoing status as follows.
(1) SEP event occurrence prediction using a classification task with explainable AI (WideLeaning)
(2) SEP event particle flux, fluence, and time-series forecasting using a regression task with the machine and deep learning methods
We discuss possible SEP event prediction schemes with these developed models and the future prospects of this industry-academia collaboration.