CAREER: From Data to Knowledge and Decisions for Global-Scale Ecological Sustainability

University Of Massachusetts Amherst, Amherst MA

Investigators

Abstract

Emerging global data resources---for example, from citizen science projects, animal tracking devices, and earth observation instruments---hold exceptional promise for monitoring biodiversity, advancing scientific discovery, and guiding decisions to conserve Earth's natural systems. However, the full potential of these novel data resources has not been realized. The data is heterogeneous, high-dimensional, and spatiotemporal. Planning problems for conservation and sustainable development have huge numbers of states and actions, and significant modeling uncertainty. This career-development project supports an integrated program of research, teaching, and outreach to develop new algorithms to convert data into knowledge and decisions for global-scale ecological sustainability. The methods are expected to help improve scientific understanding of animal populations, increase the value of citizen science data, and contribute to development plans that balance social, economic, and ecological objectives. The technical research goals are to develop new models and algorithms for reasoning about large-scale probabilistic models and for network-based spatiotemporal planning. A new class of graphical models based on probability generating functions is proposed to cleanly model animal populations, and to provide a new and flexible class of deep generative models for general multivariate count data. A novel inference approach based on automatic differentiation will be combined with existing inference techniques to support scalable inference and learning. Algorithms that utilize causal reasoning will be developed to tease apart process from noise in citizen science data. Algorithms that exploit hierarchical structure in spatiotemporal domains will be developed to optimize multiple objectives in a highly scalable way to support sustainable development. The education plan will promote diversity, train students to conduct computer science research to solve societally relevant problems, and improve student learning outcomes through a combination of research, instructional, and mentoring activities. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.