Request for Machine Learning Tutorials
The ESIP Lab facilitates opportunities for the science community to increase their technical savvy
while increasing our collective knowledge of Earth system science.
$5,000 – $7,000
Full Proposal Deadline:
May 16, 2022 at 11:00 p.m. PDT
This Request for Proposals (RPF) seeks individuals interested in furthering the impact of their science by transitioning their machine learning (ML) workflows into well-documented, interactive, and shareable tutorials. Specifically, this RFP seeks tutorials related to hydrology, seismology, and/or the cryosphere. A full-stack ML workflow should include:
- Pre-processing: open data → ML-ready training format
- ML: training, validation, and testing
- Post-processing: ML output → information products
Tutorials must be open-source; Python (3.8) is strongly encouraged.
An additional $2,000 of funding is available for those who use Geoweaver to compose their full-stack workflow. Geoweaver is an ESIP Lab and NASA ACCESS project that eases the management, sharing, replication, and reuse of ML workflows. Sign up for an informational webinar if you are interested in using Geoweaver for your tutorial development.
This RFP seeks to address the limited availability of educational material in ML for geoscientific applications. As such, tutorials developed through this RFP will be vetted and included in the GeoSMART curriculum, an NSF-funded initiative to educate the next generation of researchers to use and adopt powerful machine learning tools. The GeoSMART curriculum will include fundamental open-source ML toolkits and data science skills.
All accepted proposal teams will belong to a tutorial developer cohort. ESIP will facilitate cohort communication, including a dedicated #slack channel for collaboration/questions. Investigators of the NSF-funded GeoSMART project will provide mentorship to the cohort, and will provide a pathway for their tutorials to be included in future machine learning workshops/hackathons. The GeoSMART team includes geoscientists with domain-relevant expertise as well as ML specialists to oversee the development of training materials.
Tutorials should be developed at the graduate-school level, aimed at those with an interest in learning ML and Python for geoscience; tutorials should provide sufficient detail about the domain science and ML aspects of the use case presented.
Up to $5,000 ($7,000 if including Geoweaver) can be awarded for each proposal. Multiple projects will be selected for funding with the goal of making awards to as many of the highest-rated proposals as possible. No part of this award can be attributed to indirect costs.
- The Principal Investigator has not been funded through ESIP in the past year.
- The Principal Investigator (and anyone else receiving funds) must be authorized to work in the United States.
- Civil servants are eligible to serve as a PI or Co-I but are restricted from receiving ESIP funds.
Proposals will be judged on several criteria with the understanding that the proposal does not need to be equally strong in all categories to be judged likely to have an impact.
- Overall Impact: Likelihood that the tutorials developed will have broad use amongst the scientific community.
- Significance/Importance: Does the proposed tutorial address an important problem or a critical barrier to progress in the field?
- Approach: Are the overall strategy, methodology, and analyses well-reasoned? Are potential problems, alternative strategies, and benchmarks for success presented?
- Collaboration: Is the proposal collaborative between departments, institutions, or agencies?
- Budget: Does the budget reflect the appropriate use of resources to complete the tutorial in a timeline consistent with what has been outlined in the proposal?
Projects chosen for an award will be expected to do the following:
- Tutorials should be full-stack and written in Python (recommended 3.8+).
- Tutorials must be able to run on an open platform (Google Colab, Microsoft Planetary Computer, Binder, etc.).
- All code must be linked to the GeoSMART GitHub page.
- All project pages and associated code must include a permissive license.
How to Apply
- Please use this template for your proposal; proposals should be no longer than 4 pages.
- Applications must be submitted electronically as a single PDF to firstname.lastname@example.org.
- Proposals must be submitted up to May 15, 2022, 11 pm PDT. ESIP will alert awardees by late June 2022.
- NOTE: Funded proposals are open and will be published on the ESIP Lab website.
For questions regarding eligibility or proposal scope, please contact Annie Burgess, ESIP Lab Director, at email@example.com.