US 8,620,837 B2Granted

Determination of a basis for a new domain model based on a plurality of learned models

AssigneeAccenture Global Services Limited

InventorsRayid Ghani, Marko Krema

Life of this patent

Estimating the term…

Loading patent…

Abstract

In a machine learning system in which a plurality of learned models, each corresponding to a unique domain, already exist, new domain input for training a new domain model may be provided. Statistical characteristics of features in the new domain input are first determined. The resulting new domain statistical characteristics are then compared with statistical characteristics of features in prior input previously provided for training at least some of the plurality of learned models. Thereafter, at least one learned model of the plurality of learned models is identified as the basis for the new domain model when the new domain input statistical characteristics compare favorably with the statistical characteristics of the features in the prior input corresponding to the at least one learned model.

Claims29

Dependent claims

A method comprising:

receiving, by a processing device, data associated with a domain input;
determining, by the processing device, statistical characteristics associated with one or more features of the domain input to generate feature statistical characteristics;
identifying, by the processing device and based on the generated feature statistical characteristics, a plurality of learned models,
each of the plurality of learned models being associated with a respective other domain input,
each respective other domain input being different than the domain input;
comparing, by the processing device, the generated feature statistical characteristics with respective statistical characteristics associated with one or more features of the plurality of learned models;
identifying, by the processing device and based on the comparing, at least one learned model of the plurality of learned models that satisfies a threshold value;
modifying, by the processing device, the identified at least one learned model to create at least one trained model,
the at least one learned model being modified by analyzing the respective statistical characteristics associated with the one or more features of the identified at least one learned model and modifying a portion of the respective statistical characteristics associated with the one or more features of the identified at least one learned model that corresponds to the generated feature statistical characteristics associated with the domain input, and
the at least one trained model being different than each of the at least one learned model;
and
using, by the processing device, the at least one trained model as a domain model to process the received data.

9 more elements

An apparatus comprising:

a processor;
and
memory including instructions that, when executed by the processor, cause the processor to:
receive data associated with a domain input;
determine statistical characteristics associated with one or more features of the domain input to generate feature statistical characteristics;
identify, based on the generated feature statistical characteristics, a plurality of learned models,
each of the plurality of learned models being associated with a respective other domain input,
each respective other domain input being different than the domain input;
compare the generated feature statistical characteristics with respective statistical characteristics associated with one or more features of the plurality of learned models;
identify, based on the comparison, at least one learned model of the plurality of learned models that satisfies a threshold value;
modify the identified at least one learned model to create at least one trained model,
the at least one trained model being different than each of the at least one learned model, and
the processor, when modifying the identified at least one learned model, is further to:
analyze the respective statistical characteristics associated with the one or more features of the identified at least one learned model;
and
modify a portion of the respective statistical characteristics associated with the one or more features of the identified at least one learned model that corresponds to the generated feature statistical characteristics associated with the domain input;
and
use the at least one trained model as a domain model to process the received data.

15 more elements

A non-transitory computer readable medium storing instructions, the instructions comprising:

one or more instructions which, when executed by at least one processor, cause the at least one processor to:
receive data associated with a domain input;
determine statistical characteristics associated with one or more features of the domain input to generate feature statistical characteristics;
identify, based on the generated feature statistical characteristics, a plurality of learned models,
each of the plurality of learned models being associated with a respective other domain input,
each respective other domain input being different than the domain input;
compare the generated feature statistical characteristics with respective statistical characteristics associated with one or more features of the plurality of learned models;
modify the identified at least one learned model to create at least one trained model,
the at least one trained model being different than each of the at least one learned model, and
the one or more instructions to modify the at least one learned model include:
one or more instructions to analyze the respective statistical characteristics associated with the one or more features of the identified at least one learned model;
and
one or more instructions to modify a portion of the respective statistical characteristics associated with the one or more features of the identified at least one learned model that corresponds to the generated feature statistical characteristics associated with the domain input;
and
use the at least one trained model as a domain model to process the received data.

12 more elements

Dependencies are inferred from the claim text.