System and Method of Decoding Supply Chain Signatures

Abstract

A system and method for automated machine learning supply chain planning having a computer with a processor and memory and configured to receive a first supply chain network model having one or more material constraints for operations of a first supply chain network. Embodiments include transforming the first supply chain network model into a digital image, training an auto-encoder model to reduce the dimensionality of an input vector, and locating one or more items in the first supply chain network.

What is claimed is: 
     
         1 . A computer-implemented method for automated machine learning supply chain planning, comprising:
 receiving, by a computer comprising a processor and a memory, a first supply chain network model comprising an object model of a first supply chain network;   transforming, by the computer, the first supply chain network model expressed as an integer array into a digital image;   training, by the computer, an auto-encoder model to reduce a dimensionality of an input vector corresponding to the digital image by iteratively adjusting weights of one or more layers of the auto-encoder model using a back propagation of error until the auto-encoder model reconstructs the digital image from a supply chain signature; and   locating, by the computer, one or more items in the first supply chain network based at least in part on an item location combination matrix.   
     
     
         2 . The computer-implemented method of  claim 1 , wherein the received object model further comprises binary data. 
     
     
         3 . The computer-implemented method of  claim 1 , further comprising:
 transforming, by the computer, the object model from a binary string to an integer array,   wherein each byte of the binary string corresponds to an integer of the integer array.   
     
     
         4 . The computer-implemented method of  claim 1 , wherein a size of the digital image is based at least in part on a size of the integer array. 
     
     
         5 . The computer-implemented method of  claim 1 , wherein the supply chain signature identifies values of four-dimensional vectors representing static and dynamic properties of the first supply chain network. 
     
     
         6 . The computer-implemented method of  claim 1 , wherein the back propagation of error minimizes differences between the digital image and an output image. 
     
     
         7 . The computer-implemented method of  claim 6 , wherein the differences are calculated based on calculating a root mean-squared error. 
     
     
         8 . A system of automated machine learning supply chain planning, comprising:
 a computer, comprising a processor and memory, the computer configured to:
 receive a first supply chain network model comprising an object model of a first supply chain network; 
 transform the first supply chain network model expressed as an integer array into a digital image; 
 train an auto-encoder model to reduce a dimensionality of an input vector corresponding to the digital image by iteratively adjusting weights of one or more layers of the auto-encoder model using a back propagation of error until the auto-encoder model reconstructs the digital image from a supply chain signature; and 
 locate one or more items in the first supply chain network based at least in part on an item location combination matrix. 
   
     
     
         9 . The system of  claim 8 , wherein the received object model further comprises binary data. 
     
     
         10 . The system of  claim 8 , wherein the computer is further configured to:
 transform the object model from a binary string to an integer array, wherein each byte of the binary string corresponds to an integer of the integer array.   
     
     
         11 . The system of  claim 8 , wherein a size of the digital image is based at least in part on a size of the integer array. 
     
     
         12 . The system of  claim 8 , wherein the supply chain signature identifies values of four-dimensional vectors representing static and dynamic properties of the first supply chain network. 
     
     
         13 . The system of  claim 8 , wherein the back propagation of error minimizes differences between the digital image and an output image. 
     
     
         14 . The system of  claim 13 , wherein the differences are calculated based on calculating a root mean-squared error. 
     
     
         15 . A non-transitory computer-readable medium embodied with software, the software when executed:
 receives a first supply chain network model comprising an object model of a first supply chain network;   transforms the first supply chain network model expressed as an integer array into a digital image;   trains an auto-encoder model to reduce a dimensionality of an input vector corresponding to the digital image by iteratively adjusting weights of one or more layers of the auto-encoder model using a back propagation of error until the auto-encoder model reconstructs the digital image from a supply chain signature; and   locates one or more items in the first supply chain network based at least in part on an item location combination matrix.   
     
     
         16 . The non-transitory computer-readable medium of  claim 15 , wherein the received object model further comprises binary data. 
     
     
         17 . The non-transitory computer-readable medium of  claim 15 , wherein the software when executed further:
 transforms the object model from a binary string to an integer array, wherein each byte of the binary string corresponds to an integer of the integer array.   
     
     
         18 . The non-transitory computer-readable medium of  claim 15 , wherein a size of the digital image is based at least in part on a size of the integer array. 
     
     
         19 . The non-transitory computer-readable medium of  claim 15 , wherein the supply chain signature identifies values of four-dimensional vectors representing static and dynamic properties of the first supply chain network. 
     
     
         20 . The non-transitory computer-readable medium of  claim 15 , wherein the back propagation of error minimizes differences between the digital image and an output image, and wherein the differences are calculated based on calculating a root mean-squared error.