Back to results
US 11,900,259 B2Granted
Systems and methods for parameter optimization
AssigneeKinaxis Inc.
InventorsSebastien Ouellet, Phillip Williams, Nathaniel Stanley, Jeffery Downing, Liam Hebert
Life of this patent
Estimating the term…
Abstract
Methods and systems that provide one or more recommended configurations to planners using large data sets in an efficient manner. These methods and systems provide optimization of objectives using a genetic algorithm that can provide parameter recommendations that optimize one or more objectives in an efficient and timely manner. The methods and systems disclosed herein are flexible enough to satisfy diverse use cases.
Claims15
A computer-implemented method for data configuration of a data set for use in a machine learning system of a supply chain system, the method comprising the steps of:
- defining (i) one or more objectives and (ii) one or more supply chain parameters for evaluating each of the one or more objectives;
- generating randomly an initial set of parameters to produce an initial configuration of the data set, wherein generating the initial set of parameters comprises:
- generating a tree structure of the one or more supply chain parameters, the tree structure comprising a plurality of leaf nodes and one or more node levels;
- and
- generating an initial population of trees;
- evaluating, a fitness function of each of the one or more objectives at each of the plurality of leaf nodes of the population of trees;
- obtaining an initial Pareto Front comprising the leaf nodes that are non-dominated by other leaf nodes;
- applying recursively a genetic algorithm to each of the leaf nodes that form the initial Pareto Front, thereby generating new sets of leaf nodes forming one or more hybrid Pareto fronts, until the initial Pareto Front converges with a final Pareto Front forming a converged Pareto Front;
- generating recommended configurations for the data set based on the leaf nodes on the converged Pareto Front, wherein the leaf nodes on the converged Pareto Front represent the objectives associated with the supply chain parameters;
- selecting one or more of the recommended configurations and configuring the data set using the one or more recommended configurations for use in the machine learning system of the supply chain system;
- and
- operating the machine learning system of the supply chain system based on the configured data set.
9 more elements
A system for data configuration of a data set for use in a machine learning system of a supply chain system, the system comprising:
- a processor;
- and
- a memory storing instructions that, when executed by the processor, configure the system to:
- define (i) one or more objectives and (ii) one or more supply chain parameters for evaluating each of the one or more objectives;
- generate randomly an initial set of parameters to produce an initial configuration of the data set, wherein generating the initial set of parameters comprises:
- generating a tree structure of the one or more supply chain parameters, the tree structure comprising a plurality of leaf nodes and one or more node levels;
- and
- generating an initial population of trees;
- evaluate a fitness function of each of the one or more objectives at each of the plurality of leaf nodes of the population of trees;
- obtain an initial Pareto Front comprising the leaf nodes that are non-dominated by other leaf nodes;
- apply recursively, a genetic algorithm to each of the leaf nodes that form the initial Pareto Front, thereby generating new sets of leaf nodes forming one or more hybrid Pareto fronts, until the initial Pareto Front converges with a final Pareto Front forming a converged Pareto Front;
- generate recommended configurations for the data set based on the leaf nodes on the converged Pareto Front, wherein the leaf nodes on the converged Pareto Front represent the objectives associated with the supply chain parameters;
- select one or more of the recommended configurations and configuring the data set using the one or more recommended configurations for use in the machine learning system of the supply chain system;
- and,
- operate the machine learning system of the supply chain system based on the configured data set.
12 more elements
A non-transitory computer-readable storage medium for data configuration of a data set for use in a machine learning system of a supply chain system, the computer-readable storage medium including instructions that when executed by a computer, cause the computer to:
- define (i) one or more objectives and (ii) one or more supply chain parameters for evaluating each of the one or more objectives;
- generate randomly an initial set of parameters to produce an initial configuration of the data set, wherein generating the initial set of parameters comprises:
- generating a tree structure of the one or more supply chain parameters, the tree structure comprising a plurality of leaf nodes and one or more node levels;
- and
- generating an initial population of trees;
- evaluate a fitness function of each of the one or more objectives at each of the plurality of leaf nodes of the population of trees;
- obtain an initial Pareto Front comprising the leaf nodes that are non-dominated by other leaf nodes;
- apply recursively a genetic algorithm to each of the leaf nodes that form the initial Pareto Front, thereby generating new sets of leaf nodes forming one or more hybrid Pareto fronts, until the initial Pareto Front converges with a final Pareto Front forming a converged Pareto Front;
- generate recommended configurations for the data set based on the leaf nodes on the converged Pareto Front, wherein the leaf nodes on the converged Pareto Front represent the objectives associated with the supply chain parameters;
- select one or more of the recommended configurations and configuring the data set using the one or more recommended configurations for use in the machine learning system of the supply chain system;
- and
- operate the machine learning system of the supply chain system based on the configured data set.
9 more elements
Dependencies are inferred from the claim text.