Parallel Development and Deployment for Machine Learning Models

摘要

Example systems and methods of developing a learning model are presented. In one example, a sample data set to train a first learning algorithm is accessed. A number of states for each input of the sample data set is determined A subset of the inputs is selected, and the sample data set is partitioned into a number of partitions equal to a combined number of states of the selected inputs. A second learning algorithm is created for each of the partitions, wherein each second learning algorithm receives the unselected inputs. Each of the second learning algorithms is assigned to a processor and trained using the samples of the partition corresponding to that algorithm. Decision logic is generated to direct each of a plurality of operational data units as input to one of the second learning algorithms based on states of the selected inputs of the operational data unit.

主权项

1. A method of developing a learning model, the method comprising:
accessing a sample data set to train the learning model, wherein the learning model comprises a first learning algorithm having a number of inputs and a number of outputs, and wherein each sample of the sample data set comprises a value for each of the inputs and the outputs; determining a number of states for each of the inputs based on the sample data set; selecting a subset of the inputs; partitioning the sample data set into a number of partitions equal to a combined number of states of the selected inputs, wherein each sample of a partition exhibits a state of the selected inputs corresponding to the partition; creating a second learning algorithm for each of the partitions, wherein the second learning algorithm of a corresponding partition comprises logic of the first learning algorithm in which the state of the selected inputs corresponds to the partition, and wherein the second learning algorithm is configured to receive as input those of the inputs that are not the selected inputs; assigning each of the second learning algorithms to one of a plurality of processors of a computing system; training each of the second learning algorithms on the processor assigned to the second learning algorithm using the samples of the partition corresponding to the second learning algorithm; and generating decision logic configured to direct each of a plurality of operational data units as input to one of the second learning algorithms, wherein each of the operational data units comprises a value for each of a plurality of inputs corresponding to the inputs of the sample data set, and wherein the directing of the operational data units is based on a state of the selected inputs corresponding to the operational data unit.