| CPC G06Q 10/087 (2013.01) [G06N 20/00 (2019.01)] | 20 Claims |
|
1. A method comprising:
applying, by one or more processors, a machine learning model to a dataset to predict distributions of a plurality of inventory variables having future uncertainty, wherein the plurality of inventory variables are associated with a supply chain comprising a plurality of facilities;
generating, by the one or more processors, a graphical network model of the supply chain, wherein the graphical network model comprises a plurality of nodes corresponding to the plurality of facilities of the supply chain and a plurality of edges, wherein a particular edge of the plurality of edges indicates a facility of the plurality of facilities that supplies at least one other facility of the plurality of facilities;
iteratively determining, by the one or more processors, local distributions of the inventory variables for each node of the graphical network model, wherein each local distribution of the inventory variables corresponds to a distribution of the inventory variables at a particular node of the plurality of nodes of the graphical network model, and wherein the local distributions are propagated using iterative message passing between the nodes, where each node sends its local distribution to connected nodes and updates its local distribution based on distributions received from connected nodes during each iteration until a stop criterion is met; and
determining, by the one or more processors, an inventory management recommendation for at least one facility based on the local distribution of the inventory variables subsequent to the stop criterion being met.
|