Sequential Decision Making in Computational Sustainability Through Adaptive Submodularity

8 AI MAGAZINE One of the central challenges in computational sustainability is how to allocate resources in order to optimize long-term objectives. An archetypal application is conservation planning: managers recommend patches of land in order to achieve long-term conservation of biodiversity. In this and similar applications, we typically have to make decisions over time: financial resources (or other budgets) are periodically made available and should be used effectively. For example, every year, a certain budget may be available to support land conservation. The problem of how to optimally use this budget over time, facing uncertainty about the availability of future resources, is a challenging optimization problem. Many other decisions have to be made under substantial uncertainty about ecological function in the system of interest. Often times, this uncertainty can be partially reduced by gathering information, for example, through the application of management actions coupled with monitoring of system responses, or through other studies or experiments, allowing for improved management outcomes. Acquiring such information, however, is usually expensive. Thus, it becomes an important and challenging task to obtain the most valuable (decision relevant) information at minimum cost. In general, sequential decision making under uncertainty