Wide and deep machine learning models

摘要

The system includes one or more computers and one or more storage devices storing instructions that, when executed by the computer, cause the computer to implement a combined machine learning model for processing input to generate predictive output. The combined model comprises: a deep machine learning model that processes features to generate a deep model output; A wide area machine learning model that processes the features to generate a wide area model output; And a combination layer that processes the deep model output generated by the deep machine learning model and the wide area model output generated by the wide area machine learning model to generate a prediction output. By including both the deep machine learning model and the wide area machine learning model, the combined machine learning dodel can obtain both memorization and generalization advantages and thus perform better when predicting output from input features.