ARS AI Innovation Fund
ARS is establishing an Artificial Intelligence Center of Excellence (AICOE). The center will enable ARS science by enhancing the adoption and use of artificial intelligence (AI) and machine learning (ML) tools and methods in Agricultural research.
This year the ACIOE is holding a call for proposals to support the 2021 AI Innovation Fund.
For funding by the AI Innovation Fund we seek projects that either:
- develop an AI/ML method that empowers many ARS scientists to answer a specific question or problem, or
- Develop or adapt AI/ML to create a prototype digital product that solves a need for producers or agricultural researchers, or
- develop or adapt an AI/ML method to test a specific hypothesis or answer a specific problem of agricultural importance
Researchers developing a method or digital product are encouraged to define a minimum viable product as a deliverable. All proposals are encouraged to utilize of SCINet computing resources including the high-performance computing clusters, Ceres and Atlas equipped with graphical processing units (GPU), commodity, and high memory nodes. The systems support containerized workflows. A SCINet account will be needed to access these resources; see the SCINet web page to apply (scinet.usda.gov).
The proposal should be primarily focused on developing, adapting, and applying statistical and optimization methods that fall into the category of AI or ML. AI methods involve automated decision making from data and use methods in the subfields:
- machine learning
- mathematical optimization (integer programming and operations research)
- logic programming (not normal computer programming)
- Machine learning involves learning a model from data and making decisions. ML methods may include:
- Tasks like classification, regression, dimensionality reduction, clustering
- Domain areas like natural language processing, image segmentation, image classification
- Statistical methods like Gaussian processes, neural networks, Bayesian networks, support vector machines
Successful projects should address real-world ML model concerns like dataset shift and have AI/ML development be a primary focus, and NOT just use an ML methods like principal components analysis as part of their research. Educational activities are not supported by this call, but those interested in education and training support should see the Training page on the SCINet web page.
We expect to support 4 to 6 projects with a maximum budget request of $100,000 per project. Funds must be spent in the fiscal year 2021 (before 9/30/2021) which may require an agreement with a university partner or the Oak Ridge Institute for Science and Education (ORISE).
The proposal should be a maximum of 2 pages in length, not including references,. The proposal should describe a specific challenge, method, or tool to be developed or applied to solve a problem/challenge or to answer an agricultural question. Deliverables for the project should be defined. A budget and budget explanation of no more than 1 page should be included in addition to the two-page proposal. All pages should be single-spaced in Times New Roman 11-point font with 1-inch margins.
5 PM Eastern, May 25, 2021; successful proposals will be contacted as soon as possible to allow a start date in June.
USDA-ARS Category 1, 4, or 6 scientists with supervisor approval.
To submit please create a single PDF with these documents:
Email to: ARS-SCINet-CSIO@usda.gov with the subject line “AI Innovation application”