Speaker
Description
The CORonal mass ejection, solar eNERgetic particle and flare forecaSTing from phOtospheric sigNaturEs (CORNERSTONE) project focuses on the prediction of intense solar events through the application of machine learning (ML) techniques to real observational data. This domain poses significant challenges to reproducibility, primarily due to the heterogeneous nature of the data and the complexity of applying ML methods in a physically meaningful way. Real-world solar data are typically acquired by diverse teams of specialists - from electronics and engineering to physics - and usually originate from in-flight instruments. As a result, these data are subject to a variety of issues not directly related to the measurements themselves, but to the conditions and limitations under which data acquisition occurs. Moreover, from a methodological perspective, while ML algorithms are widely accessible, their rigorous and interpretable use in the context of solar physics requires careful data preprocessing. In particular, ensuring that models are trained on stratified and statistically consistent distributions is critical for producing reliable results, especially when moving from offline experimentation to real-time operational forecasting. The CORNERSTONE project aims to address these issues by developing methodologies and tools that enable reproducible and verifiable ML-based research. It provides high-quality criteria for data stratification, the construction of curated datasets for benchmarking, and a framework for testing multiple ML algorithms under reproducible conditions. The ultimate objective of CORNERSTONE is the development of an open-source Python library that integrates data preprocessing, data verification, model implementation, and validation tools. This library will facilitate reproducibility by enabling researchers to both verify their own models and independently assess those developed by others, fostering transparency and collaboration in the solar forecasting research community.
| Do you plan to attend in-person or online? | In-person |
|---|
