ROSES 2022: Heliophysics Artificial Intelligence/Machine Learning-Ready Data
Funding Agency:
- NASA
This Heliophysics artificial intelligence/machine learning (AI/ML)-Ready Data (H-ARD) Program element solicits proposals that enable the advancement of the goals and objectives of NASA Heliophysics by developing new tools and methods for the generation of AI/ML-ready datasets from existing research and mission data. H-ARD is a component of the Heliophysics Research Program and proposers interested in this program element should read B.1, the Heliophysics Research Program Overview for Heliophysics-specific requirements. Common requirements for all ROSES elements and proposals are found in the ROSES Summary of Solicitation and 35TUNASA Guidebook for ProposersU35T and the order of precedence for proposers is the following: ROSES Element B.16 (this document) takes precedence followed by B.1, The Heliophysics Research Program Overview, followed by the ROSES-2022 Summary of Solicitation and, finally, the 2022 NASA Proposer's Guidebook. Proposers should be familiar with all of these resources. Artificial intelligence (AI) and its subset, machine learning (ML), have become potentially effective means for achieving scientific goals and collecting and analyzing large data sets. Scientists have begun to use "theory-aided" or "knowledge-aided" AI to achieve breakthroughs. These tools and techniques can lead the way to a new understanding and drive science concepts for future strategic missions and other research efforts. The AI/ML approach is data-intensive, with data sets used for training, tuning and testing AI/ML processes. Heliophysics data must be AI/ML- ready to be able to apply various methods and tools and to be stored as AI/ML catalogs and archives for public use. This process includes creating “clean” datasets which might require fixing structural errors, handling missing data, removing non-physical outlier points, and/or filtering observations. With a clean dataset, even simple algorithms can yield important insights. The quality of the data is key to the quality of machine learning algorithms developed from the data. In creating such a dataset, specific domain expertise or collaboration with the persons responsible for the data quality (for example, the PI of a satellite instrument) becomes very important.
TBD
Open until Jan 18, 2023
Katya Verner Heliophysics Division Science Mission Directorate NASA Headquarters Washington, DC 20546-0001 Telephone: (202) 358-1213 Email: ekaterina.m.verner@nasa.gov