Changes New model type: Transaction Fraud Insights, which is optimized for online transaction fraud. Stored Events, which allows customers to send and store data directly within Amazon Fraud Detector. Batch Import, which allows customers to upload a CSV file of historic event data for processing and storage
Changes Enhanced GetEventPrediction API response to include risk scores from imported SageMaker models
Changes Updated an element of the DescribeModelVersion API response (LogitMetrics -> logOddsMetrics) for clarity. Added new exceptions to several APIs to protect against unlikely scenarios.
Changes This release adds support for ML Explainability to display model variable importance value in Amazon Fraud Detector.
Changes This release adds support for Batch Predictions in Amazon Fraud Detector.
Changes Added support for cancelling a model version that is TRAINING_IN_PROGRESS.
Changes Added support for deleting resources like Variables, ExternalModels, Outcomes, Models, ModelVersions, Labels, EventTypes and EntityTypes. Updated DeleteEvent operation to catch missing exceptions.
Changes Moved the eventTypeName attribute for PutExternalModel API to inputConfiguration. Model ID's no longer allow hyphens.
Changes GetPrediction has been replaced with GetEventPrediction. PutExternalModel has been simplified to accept a role ARN.
Changes Introduced flexible model training dataset requirements for Online Fraud Insights so that customers can choose any two inputs to train a model instead of being required to use 'email' and 'IP address' at minimum. Added support for resource ARNs, resource tags, resource-based IAM policies and identity-based policies that limit access to a resource based on tags. Added support for customer-managed customer master key (CMK) data encryption. Added new Event Type, Entity Type, and Label APIs. An event type defines the structure for an event sent to Amazon Fraud Detector, including the variables sent as part of the event, the entity performing the event, and the labels that classify the event. Introduced the GetEventPrediction API.
Changes Added support for a new rule engine execution mode. Customers will be able to configure their detector versions to evaluate all rules and return outcomes from all 'matched' rules in the GetPrediction API response. Added support for deleting Detectors (DeleteDetector) and Rule Versions (DeleteRuleVersion).
Changes Amazon Fraud Detector is a fully managed service that makes it easy to identify potentially fraudulent online activities such as online payment fraud and the creation of fake accounts. Amazon Fraud Detector uses your data, machine learning (ML), and more than 20 years of fraud detection expertise from Amazon to automatically identify potentially fraudulent online activity so you can catch more fraud faster.