Interpretable rule generation using loss preserving transformation
AssigneeInternational Business Machines Corporation
InventorsAmit Dhurandhar, Sechan Oh, Marek Petrik
Life of this patent
Estimating the term…
Abstract
Techniques facilitating interpretable rule generation using loss-preserving transformation are provided. In one example, a computer-implemented method can comprise evaluating, by a system operatively coupled to a processor, an input data set that comprises three data categories. The computer-implemented method can also comprise transforming, by the system, the input data set into a transformed data set. The transformed data set can comprise two data categories determined based on the three data categories. Transforming the input data set can comprise determining a first cost associated with the transformed data set is no greater than a second cost associated with the input data set.
Claims20
A system, comprising:
- a memory that stores computer executable components;
- and
- a processor that executes the computer executable components stored in the memory, wherein the computer executable components comprise:
- an evaluation component that, employs the processor, to evaluate an input data set associated with a user, wherein the input data set comprises three related data inputs, wherein the three related data inputs comprise a first set of data input comprising features describing known information about the user, a second set of data input comprising historical recommended actions provided to the user, and a third set of data input comprising outcomes describing how the user responded to the historical recommended actions;
- and
- a conversion component that, employs the processor to:
- generate a prediction model based on the three related data inputs and based on a reduction of a 0-1 loss function for a misclassification cost associated with the three related data inputs, wherein the prediction model transforms the three related data inputs into a transformed data set comprising output data pairs respectively comprising one or more features and one or more recommended actions that satisfy a defined condition related to a conversion probability of achieving one or more outcomes with respect to the one or more recommended actions, and wherein the 0-1 loss function is based on a first misclassification cost associated with the transformed data set being no greater than a second misclassification cost associated with the input data set, wherein the generation of the prediction model comprises generation of an estimator function that meets a conversion rate for recommended actions for features to outcomes that meets the conversion probability, and the estimator function applies respective weights to the output data pairs to meet the conversion probability, and
- generate, using the prediction model, at least one recommended action associated with content based on the transformed data set.
A computer program product for facilitating recommendation generation, the computer program product comprising a non-transitory computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor to cause the processor to:
- evaluate, by the processor, an input data set associated with a user, wherein the input data set comprises three related data inputs, wherein the three related data inputs comprise a first set of data input comprising features describing known information about the user, a second set of data input comprising historical recommended actions provided to the user, and a third set of data input comprising outcomes describing how the user responded to the historical recommended actions;
- generate, by the processor, a prediction model based on the three related data inputs and based on a reduction of a 0-1 loss function for a misclassification cost associated with the three related data inputs, wherein the prediction model transforms the three related data inputs into a transformed data set comprising output data pairs respectively comprising one or more features and one or more recommended actions that satisfy a defined condition related to a conversion probability of achieving one or more outcomes with respect to the one or more recommended actions, and wherein the 0-1 loss function is based on a first misclassification cost associated with the transformed data set being no greater than a second misclassification cost associated with the input data set, wherein the generation of the prediction model comprises generation of an estimator function that meets a conversion rate for recommended actions for features to outcomes that meets the conversion probability, and the estimator function applies respective weights to the output data pairs to meet the conversion probability, and
- generate, by the processor using the prediction model, at least one recommended action based on the transformed data set.
Dependencies are inferred from the claim text.