Unlock the Power of Personalization: How to Choose the Perfect HRNN Recipes with AutoML in Amazon Personalize
Optimizing Amazon Personalize with AutoML
Amazon Personalize offers a powerful feature called AutoML, which automatically selects the most suitable Hierarchical Recurrent Neural Network (HRNN) recipe based on the input data analysis. This feature can be enabled by setting the performAutoML parameter to true when calling the CreateSolution API.
Alternatively, you can specify a list of recipes for Amazon Personalize to examine and determine the best recipe based on a specified metric. To do this, call the CreateSolution operation, set performAutoML to true, omit the recipeArn parameter, include the solutionConfig parameter, and specify the metricName and recipeList as part of the autoMLConfig object.
Recipe Selection Options
| performAutoML | recipeArn | solutionConfig | Result |
|---|---|---|---|
| true | Ignored | Ignored | Amazon Personalize selects the recipe for the optimization metric from the list |
| true | Ignored | autoMLConfig: metricName and recipeList specified | Amazon Personalize selects the recipe for the optimization metric from the list |
| Ignored | You specify recipe | Ignored | You specify recipe and override default training attributes |
When performAutoML is set to true, all parameters of the solutionConfig object are ignored except autoMLConfig.