Introducing automated training for solutions with Amazon Personalize

Amazon Personalize is pleased to announce automated training for our solution. Training a solution is fundamental to maintaining model effectiveness and ensuring that recommendations match users’ evolving behaviors and preferences. Patterns and trends in data change over time, so retraining your solution with the latest relevant data allows your model to learn and adapt, improving predictive accuracy. Automatic training generates new solution versions to reduce model drift and keep recommendations relevant to the end user’s current behavior while including the latest items. Ultimately, automated training provides a more personalized and engaging experience that adapts to changing preferences.

Amazon Personalize uses machine learning (ML) to accelerate your digital transformation, making it easy to integrate personalized recommendations into your existing websites, applications, email marketing systems, and more. Amazon Personalize allows developers to quickly implement customized personalization engines without requiring their ML expertise. Amazon Personalize provisions the necessary infrastructure and manages your entire ML pipeline, including processing your data, identifying features, using appropriate algorithms, and training, optimizing, and hosting customized models based on your data. . All data is encrypted to protect your privacy.

This post describes the process of configuring automatic training so that your solutions and recommendations remain accurate and relevant.

Solution overview

a solution Refers to the combination of an Amazon Personalize recipe, customized parameters, and one or more solution versions (trained models). When you create a custom solution, you specify a recipe that matches your use case and configure training parameters. In this post, we will configure automatic training with training parameters.

Prerequisites

To enable automatic training for your solution, you must first configure your Amazon Personalize resources. First, create dataset groups, schemas, and datasets to represent items, interactions, and user data. For instructions, see Getting Started (Console) or Getting Started (AWS CLI).

Once the data has been imported, you are ready to create your solution.

Create the solution

To set up automatic training, follow these steps:

1. Create a new solution in the Amazon Personalize console.
2. Specify a name for your solution, select the type of solution you want to create, and select a recipe.
3. Add tags as needed. For more information about tagging Amazon Personalize resources, see Tagging Amazon Personalize Resources.
4. To use automatic training, automatic training section, selection turn on Specify the training frequency.

Automatic training is enabled by default and trains once every 7 days. Depending on your business needs, he can configure the frequency of training from 1 to 30 days.

1. If the recipe generates item recommendations or user segments, optionally training column section, select the columns that Amazon Personalize considers when training versions of your solution.
2. inside Setting hyperparameters sectionoptionally configure hyperparameter options based on your recipe and business needs.
3. Provide and select additional configurations Next.
   
4. Review the solution details and confirm that automatic training is configured as expected.
5. choose Creating a solution.
   

Amazon Personalize automatically creates your first solution version.a solution version Refers to a trained ML model. When a solution version of your solution is created, Amazon Personalize trains a model that supports the solution version based on your recipe and training settings. It may take up to an hour for solution version creation to begin.

Below is sample code for creating a solution with automatic training using AWS SDKs.

import boto3 
personalize = boto3.client('personalize')

solution_config = {
    "autoTrainingConfig": {
        "schedulingExpression": "rate(3 days)"
    }
}

recipe = "arn:aws:personalize:::recipe/aws-similar-items"
name = "test_automatic_training"
response = personalize.create_solution(name=name, recipeArn=recipe_arn, datasetGroupArn=dataset_group_arn, 
                            performAutoTraining=True, solutionConfig=solution_config)

print(response['solutionArn'])
solution_arn = response['solutionArn'])

After you create a solution, you can check whether automatic training is enabled on the solution details page.

You can also verify that automatic training is enabled via the AWS SDK using the following sample code.

response = personalize.describe_solution(solutionArn=solution_arn)
print(response)

The response contains the following fields performAutoTraining and autoTrainingConfigThe value set in is displayed. CreateSolution phone.

The solution details page also displays the automatically created version of the solution.of Type of training The column specifies whether the solution version was created manually or automatically.

You can also return a list of solution versions for a specific solution using the following sample code.

response = personalize.list_solution_versions(solutionArn=solution_arn)['solutionVersions']
print("List Solution Version response\n")
for val in response:
    print(f"SolutionVersion: {val}")
    print("\n")

The response contains the fields trainingTypespecifies whether the solution version was created manually or automatically.

Once your solution version is ready, you can create a campaign for it.

Create a campaign

a motion Deploy the solution version (trained model) to generate real-time recommendations. With Amazon Personalize, you can streamline your workflow and automate the deployment of the latest solution versions to your campaigns through automatic synchronization. To set up automatic sync, follow these steps:

1. Create a new campaign in the Amazon Personalize console.
2. Specify a name for your campaign.
3. Select the solution you just created.
4. select Automatically use the latest solution version.
5. Sets the minimum provisioning transactions per second.
   
6. Create a campaign.

Your campaign is ready if the status is: ACTIVE.

Below is the sample code for creating a campaign. syncWithLatestSolutionVersion set to true Use AWS SDK.You also need to add the suffix $LATEST to solutionArn in solutionVersionArn When setting syncWithLatestSolutionVersion to true.

campaign_config = {
    "syncWithLatestSolutionVersion": True
}
resource_name = "test_campaign_sync"
solution_version_arn = "arn:aws:personalize:<region>:<accountId>:solution/<solution_name>/$LATEST"
response = personalize.create_campaign(name=resource_name, solutionVersionArn=solution_version_arn, campaignConfig=campaign_config)
campaign_arn = response['campaignArn']
print(campaign_arn)

You can check whether auto-sync is enabled for the selected campaign on the campaign details page. When enabled, campaigns are automatically updated to use the latest solution version, whether created automatically or manually.

Use the following sample code to verify the following via AWS SDKs: syncWithLatestSolutionVersion Enabled:

response = personalize.describe_campaign(campaignArn=campaign_arn)
Print(response)

The response contains the fields syncWithLatestSolutionVersion under campaignConfigThe value set is displayed. CreateCampaign phone.

After you update a campaign and create a campaign, you can enable or disable the option to automatically use the latest solution version in the Amazon Personalize console.Similarly, you can enable or disable syncWithLatestSolutionVersion and UpdateCampaign Use AWS SDK.

conclusion

With automatic training, you can reduce model drift and keep your recommendations relevant by streamlining your workflow and automating the deployment of the latest solution version with Amazon Personalize.

For more information about optimizing the user experience with Amazon Personalize, see the Amazon Personalize Developer Guide.

About the author

bacari johnson I’m a Senior Technical Product Manager for AWS AI/ML on the Amazon Personalize team. With a background in computer science and strategy, she has a passion for product innovation. In her spare time, she enjoys traveling and exploring the great outdoors.

Ajay Venkatakrishnan I’m a software development engineer on the Amazon Personalize team. In his spare time, he enjoys writing and playing soccer.

Pranesh Anubhav I’m a senior software engineer at Amazon Personalize. He is passionate about designing machine learning systems to serve large-scale customers. Outside of work, he loves soccer and is an avid Real Madrid fan.