System and Platform for Food Security at Food Banks

In a food bank environment under increasing financial stress, increased demands, poor nutritional quality, and unpredictable levels of donated foods improvement in food security of served population requires the utilization of systems thinking approach, advanced data analytics, and optimization techniques to manage inventory, forecast supplies patterns, and combat food insecurity.

In this regard, combining sophisticated supply predictive analytics with the state of the art nutrition knowledge can provide critical insights into food resources allocation processes and inform food bank managers. Such is the contribution of the present research (“Unified framework for efficient, effective, and fair resource allocation at food banks-Approximate Dynamic Programming approach”): We build an end-to-end machine learning and optimization platform to support various decision making processes at food banks.

The back-end of our platform is a flexible optimization framework based on a tractable Approximate Dynamic Programming (ADP) algorithm for resource allocation where there are multiple resources. Since our algorithm is based on systems thinking and dynamic programming, the proposed algorithm captures randomness in the system and the effectiveness and efficiency performance measures and implicitly considers the equity performance measure. To handle the large state space (due to considering multiple resources), we develop a set of basis functions that estimate the expected utility in the system to enhance the performance of the ADP. More specically, we introduce basis functions that estimate the future utility in the system and thus help the ADP decide how much resources to keep in the inventory for future use.

Collaborators: Professor Oliver Gao, Faisal Alkaabneh, Ph.D. student
Students:  Najat Alrashed, Abdullah Islam, Sawanth Prasad, Xia He, and Xueman Liang.