US 12,608,405 B2Granted

Model generation based on modular storage of training data

AssigneeCitibank, N.A.

InventorsSanakar Narayan Karasudha Patnaik, Akshay Jain

Life of this patent

Estimating the term…

Loading patent…

Abstract

The model generation platform enables generation of machine learning models based on modular storage of user-specified training data. The platform can obtain a dataset from a different heterogenous source (e.g., a structured database, an unstructured database, a semi-structured file system, manual upload of a comma-separated value file, a spreadsheet, and/or big data) through an associated application programming interface and store this data in a first storage medium. The model generation platform can obtain an indication of a portion of the dataset from a user via a user interface and determine a second storage medium for this portion of the dataset based on an associated estimated performance metric. In response to a request for generation of a machine learning model, the model generation platform can generate a machine learning model using training data comprising a subset of the portion of the dataset.

Claims20

Dependent claims

A non-transitory, computer-readable storage medium for dynamically generating a machine learning model based on dynamically-stored training data, wherein the non-transitory, computer-readable storage medium comprises instructions thereon that, when executed by at least one data processor of a system, cause the system to:

determine, by executing a server utility application for a server system, an estimated performance metric for a portion of a dataset indicating an estimated resource usage for storing the portion of the dataset;
dynamically determine a threshold metric associated with a second storage medium of the server system,
wherein the threshold metric includes a dynamically determined performance-related metric for the second storage medium comprising at least one of:
(1) an estimated read-write speed, (2) a volatility value, or (3) a remaining available storage value;
determine, by comparing the estimated performance metric with the threshold metric associated with the second storage medium, a storage medium for storage of a representation of the portion of the dataset;
in response to determining that the estimated performance metric is less than the threshold metric, generate, within the determined storage medium comprising the second storage medium, the representation of the portion of the dataset;
in response to determining that the estimated performance metric is greater than or equal to the threshold metric, generate, within the determined storage medium comprising a third storage medium that is different from the second storage medium, the representation of the portion of the dataset;
and
generate the machine learning model that is trained using training data comprising a subset of the portion of the dataset,
wherein parameters associated with the machine learning model are stored in the determined storage medium based on the estimated performance metric.

7 more elements

A system for dynamically generating a machine learning model based on dynamically-stored training data, comprising:

at least one hardware processor;
and
at least one non-transitory memory storing instructions, which, when executed by the at least one hardware processor, cause the system to:
determine, by executing a server utility application for a server system, a metric for a portion of a dataset indicating an estimated parameter for storing the portion of the dataset;
dynamically determine a threshold metric associated with a second storage medium associated with the server system,
wherein the threshold metric includes a dynamically determined performance-related metric for the second storage medium comprising at least one of:
(1) an estimated read-write speed, (2) a volatility value, or (3) a remaining available storage value;
determine, by comparing the metric with the threshold metric associated with the second storage medium, a storage medium for storage of a representation of the portion of the dataset;
in response to determining that the metric satisfies the threshold metric, configure for generation, within the determined storage medium comprising the second storage medium, the representation of the portion of the dataset;
in response to determining that the metric does not satisfy the threshold metric, generate, within the determined storage medium comprising a third storage medium that is different from the second storage medium, the representation of the portion of the dataset;
generate the machine learning model that is trained using training data comprising a subset of the portion of the dataset,
wherein parameters associated with the machine learning model are stored in the determined storage medium based on the metric.

9 more elements

A method for dynamically generating a machine learning model based on dynamically-stored training data, comprising:

determining, by executing a server utility application for a server system, a metric for a portion of a dataset indicating an estimated parameter associated with storing the portion of the dataset;
dynamically determine a threshold metric associated with a second storage medium of the server system,
wherein the threshold metric includes a dynamically determined performance-related metric for the second storage medium comprising at least one of:
(1) an estimated read-write speed, (2) a volatility value, or (3) a remaining available storage value;
determine, by comparing the metric with the threshold metric associated with the second storage medium, a storage medium for storage of a representation of the portion of the dataset;
in response to determining that the metric is less than the threshold metric, generating, within the determined storage system comprising the second storage medium, the representation of the portion of the dataset;
in response to determining that the metric is greater than or equal to the threshold metric, generate, within the determined storage medium comprising a third storage medium that is different from the second storage medium, the representation of the portion of the dataset;
and
generating the machine learning model that is trained using training data comprising a subset of the portion of the dataset,
wherein parameters associated with the machine learning model are stored in the determined storage medium based on the metric.

7 more elements

Dependencies are inferred from the claim text.