Improving biodiversity protection through artificial intelligence
Silvestro, D., Goria, S., Sterner, T., and Antonelli, A. (2022) Nature Sustainability, DOI:10.1038/s41893-022-00851-6Download open access article:
A simulated natural system
CAPTAIN uses simulations based on an individual-based spatially explicit model of biodiversity to train policies through Reinforcement Learning. The simulations can include hundreds of species and millions of individuals and tracks global and local biodiversity changes resulting from natural processes of mortality, replacement and dispersal and from changes in anthropogenic pressure and climate. Simlated systems are used to train models that can be then applied to empirical data and to becnhmark the outcome of different conservation policies and targets.