Schematic Diagram of DAGGER in CL paradigm.
The question
How can a forecasting model keep learning as the environment changes?
Space weather forecasting requires models that can operate on evolving streams of solar and geomagnetic observations. Continual learning offers a way to update these models as new data becomes available.
As a scientific researcher with Frontier Development Lab in 2024, I led the continual learning component of DAGGER++ and supported redevelopment of SHEATH for integration of real-time solar and geomagnetic data. This work contributed to the GEO-CLOAK effort on forecasting ground-level geomagnetic perturbations.
Related work
Geoeffectiveness continual learning: GEO-CLOAK
C. Pandey et al. Technical Memorandum, NASA, 2024.
Read NASA technical memorandumGEO-CLOAK: Operational machine learning tool for global geomagnetic field perturbation forecasting
B. K. Jha, C. Pandey, O. Issan, et al. AGU, 2024. Continual learning and geomagnetic forecasting.
View poster