[−][src]Crate rusoto_sagemaker_a2i_runtime
Amazon Augmented AI (Augmented AI) (Preview) is a service that adds human judgment to any machine learning application. Human reviewers can take over when an AI application can't evaluate data with a high degree of confidence.
From fraudulent bank transaction identification to document processing to image analysis, machine learning models can be trained to make decisions as well as or better than a human. Nevertheless, some decisions require contextual interpretation, such as when you need to decide whether an image is appropriate for a given audience. Content moderation guidelines are nuanced and highly dependent on context, and they vary between countries. When trying to apply AI in these situations, you can be forced to choose between "ML only" systems with unacceptably high error rates or "human only" systems that are expensive and difficult to scale, and that slow down decision making.
This API reference includes information about API actions and data types you can use to interact with Augmented AI programmatically.
You can create a flow definition against the Augmented AI API. Provide the Amazon Resource Name (ARN) of a flow definition to integrate AI service APIs, such as Textract.AnalyzeDocument and Rekognition.DetectModerationLabels. These AI services, in turn, invoke the StartHumanLoop API, which evaluates conditions under which humans will be invoked. If humans are required, Augmented AI creates a human loop. Results of human work are available asynchronously in Amazon Simple Storage Service (Amazon S3). You can use Amazon CloudWatch Events to detect human work results.
You can find additional Augmented AI API documentation in the following reference guides: Amazon Rekognition, Amazon SageMaker, and Amazon Textract.
If you're using the service, you're probably looking for SagemakerA2iRuntimeClient and SagemakerA2iRuntime.
Structs
| DeleteHumanLoopRequest | |
| DeleteHumanLoopResponse | |
| DescribeHumanLoopRequest | |
| DescribeHumanLoopResponse | |
| HumanLoopActivationReason | Contains information about why a human loop was triggered. If at least one activation reason is evaluated to be true, the human loop is activated. |
| HumanLoopActivationResults | Information about the corresponding flow definition's human loop activation condition evaluation. Null if |
| HumanLoopInputContent | An object containing the input. |
| HumanLoopOutputContent | Information about where the human output will be stored. |
| HumanLoopSummary | Summary information about the human loop. |
| HumanReviewDataAttributes | Attributes of the data specified by the customer. Use these to describe the data to be labeled. |
| ListHumanLoopsRequest | |
| ListHumanLoopsResponse | |
| SagemakerA2iRuntimeClient | A client for the Amazon Augmented AI Runtime API. |
| StartHumanLoopRequest | |
| StartHumanLoopResponse | |
| StopHumanLoopRequest | |
| StopHumanLoopResponse |
Enums
| DeleteHumanLoopError | Errors returned by DeleteHumanLoop |
| DescribeHumanLoopError | Errors returned by DescribeHumanLoop |
| ListHumanLoopsError | Errors returned by ListHumanLoops |
| StartHumanLoopError | Errors returned by StartHumanLoop |
| StopHumanLoopError | Errors returned by StopHumanLoop |
Traits
| SagemakerA2iRuntime | Trait representing the capabilities of the Amazon Augmented AI Runtime API. Amazon Augmented AI Runtime clients implement this trait. |