Model interpretation
AssigneeH2O.ai Inc.
InventorsMark Chan, Navdeep Gill, Patrick Hall
Life of this patent
Estimating the term…
Abstract
Input data associated with a machine learning model is classified into a plurality of clusters. A plurality of linear surrogate models are generated. One of the plurality of linear surrogate models corresponds to one of the plurality of clusters. A linear surrogate model is configured to output a corresponding prediction based on input data associated with a corresponding cluster. Prediction data associated with the machine learning model and prediction data associated with the plurality of linear surrogate models are outputted.
Claims20
A method, comprising:
- generating a global surrogate model that includes a linear surrogate model and one or more non-linear surrogate models to approximate an output of a machine learning model;
- receiving, via a user interface, a selection of a data point included in the linear surrogate model;
- in response to the selection of the data point included in the linear surrogate model, updating the one or more non-linear surrogate models to be specific for the selected point;
- determining that prediction data associated with the linear surrogate model correlates with predication data associated with machine learning model and prediction data associated with one of the non-linear surrogate models correlates with the prediction data associated with the machine learning model;
- and
- utilizing the linear surrogate model and the one of the non-linear surrogate models to display within the user interface an indication of one or more input features of input data associated with the machine learning model that influenced the prediction data associated with the machine learning model.
A system, comprising:
- a processor configured to:
- generate a global surrogate model that includes a linear surrogate model and one or more non-linear surrogate models to approximate an output of a machine learning model;
- receive, via a user interface, a selection of a data point included in the linear surrogate model;
- in response to the selection of the data point included in the linear surrogate model, update the one or more non-linear surrogate models to be specified for the selected point;
- determine that prediction data associated with the linear surrogate model correlates with predication data associated with machine learning model and prediction data associated with one of the non-linear surrogate models correlates with the prediction data associated with the machine learning model;
- and
- utilize the linear surrogate model and the one of the non-linear surrogate models to display within the user interface an indication of one or more input features of input data associated with the machine learning model that influenced the prediction data associated with the machine learning model;
- and
- a memory coupled to the processor and configured to provide the processor with instructions.
A computer program product, the computer program product being embodied in a non-transitory computer readable storage medium and comprising computer instructions for:
- generating a global surrogate model that includes a linear surrogate model and one or more non-linear surrogate models to approximate an output of a machine learning model;
- receiving, via a user interface, a selection of a data point included in the linear surrogate model;
- in response to the selection of the data point included in the linear surrogate model, updating the one or more non-linear surrogate models to be specific for the selected point;
- determining that prediction data associated with the linear surrogate model correlates with predication data associated with machine learning model and prediction data associated with one of the non-linear surrogate models correlates with the prediction data associated with the machine learning model;
- and
- utilizing the linear surrogate model and the one of the non-linear surrogate models to display within the user interface an indication of one or more input features of input data associated with the machine learning model that influenced the prediction data associated with the machine learning model.
Dependencies are inferred from the claim text.