There are two research units involved in the project, that will share the effort to realize the project according to their specific competences and the role assumed in the project.<\/p>\n\n\n\n
The UNIGE unit, which will lead the project, will take care of the algorithmic development and management of the project in its entirety.
The UNITOV unit, on the other hand, will have the leadership of the scientific and technological part for which specific knowledge in solar physics is fundamental.<\/p>\n\n\n\n
The two research units will cooperate starting to define an efficient strategy, and use cases, to introduce new features and signatures pertaining to solar flare activities; the team members of the two units will have to decide which are the best elementary functions to use to generate significant features for flare, CME and SEP prediction.
The main scientific contribution of this step is expected by the UNITOV unit, while the UNIGE unit will mainly contribute with the technical development of specific algorithms.
The UNIGE unit will lead the subsequent step of the project to find the relation between new features correlated with solar activity.
Indeed, this step is strongly based on the development of machine learning methods driven by physically meaningful input features.
As a result of this joint effort the project is expected to generate new equations involving measurable features starting from data and verifiable by empirical measurements.
This verification step will be led by the UNITOV unit and supported by the UNIGE unit.
The UNIGE unit will provide interpretable machine learning algorithms, whose outputs may be used as suggestions to rethink the solar eruptive mechanisms.
In this regard, the UNITOV unit will be most involved in evaluating the entire result of the project as a feedback tool for the study of the relationship between SEPs, flares, and CMEs and enable a deeper understanding of how these phenomena are related to each other.
The UNIGE team will take the lead to develop a set of algorithms able to continuously generate a combination of non linear new features and to develop a set of kernel algorithms able to verify the capability of new features to correlate with observations.<\/p>\n\n\n\n