×

Feature processing recipes for machine learning

  • US 9,886,670 B2
  • Filed: 06/30/2014
  • Issued: 02/06/2018
  • Est. Priority Date: 06/30/2014
  • Status: Active Grant
First Claim
Patent Images

1. A system, comprising:

  • one or more computing devices configured to;

    implement a network-based interface for a network-accessible machine learning service of a provider network, the network-based interface configured to receive text representations of different client recipes from a plurality of respective clients of the network-accessible machine learning service, wherein respective client recipes received via the network-based interface specify one or more transformations to be performed on respective data sets used for model training or prediction by the machine learning service, the respective text representations of client recipes comprising one or more of;

    (a) a group definitions section indicating one or more groups of variables, wherein individual ones of the one or more groups comprise a plurality of variables on which at least one common transformation operation is to be applied, (b) an assignment section defining one or more intermediate variables, (c) a dependency section indicating respective references to one or more machine learning artifacts stored in a repository, or (d) an output section indicating one or more transformation operations to be applied to at least one entity indicated in the group definitions section, the assignment section, or the dependency section;

    validate, in accordance with (a) a set of syntax rules defined by the machine learning service and (b) a set of library function definitions for transformation operation types supported by the machine learning service, a given text representation of a given client recipe received via the network-based interface;

    generate, from the given text representation, an executable representation of the given client recipe;

    store the executable representation in the repository;

    determine that the given client recipe is to be applied to a particular data set;

    verify that the particular data set meets a run-time acceptance criterion of the given client recipe; and

    execute the executable representation of the given client recipe, using one or more selected provider network resources, to apply a particular transformation operation specified in the given client recipe to the particular data set.

View all claims
  • 1 Assignment
Timeline View
Assignment View
    ×
    ×