AWS Enhances Machine Learning in AmplifyMobile Back-End AWS Amplify is a back-end framework for mobile and Web app development. It now has improved machine learning functionality. Amplify has roots dating back to 2015, when it was known as Mobile Hub. It is described as “an opinionated collection of libraries, UI components and a command line interface to build an application backend and integrate with your iOS, Android and Web apps.” Its services include storage and authentication & authorisation, APIs (GraphQL & REST), analytics and push notifications, chatbots, AR/VR, and now, improved machine-learning functionality. The Amplify Framework now includes a Predictions category. This is in addition to other pre-trained AI services that can be used by app developers without advanced AI knowledge. These include image and video analysis, personalized recommendations, virtual assistants and forecasting. Developers can create the code amplify and add predictions to configure an app to:

  • Amazon Rekognition can identify text, entities, labels, and images. You can also identify text in scanned documents using Amazon Textract to retrieve the contents of fields and information stored in tables.
  • Amazon Translate converts text into another language, Amazon Polly converts text to speech, and Amazon Transcribe converts text to speech
  • Amazon Comprehend allows you to interpret text to determine the dominant language, key phrases, sentiment, and the syntax of unstructured texts.

“It’s never been easier to add machine-learning functionalities to a Web app or mobile app,” said Danilo Poccia (AWS evangelist), who explains how to do it in a blog post on July 31. AWS announced that Web developers could use the Amplify JavaScript library and the new Predictions category to add a variety of AI/ML use cases to their Web applications, including text translation, speech-to-text generation, image recognition, speech to speech generation, text to text, and insights from the text. Mobile developers can now use the new iOS SDK for SageMaker and Android SDK to make inference requests on SageMaker-hosted custom models with an HTTPS connection. “The Android SDK supports Amazon Textract, which allows developers to extract text from scanned documents. These services are now available in the Android and iOS SDKs. You can find more information in the Predictions documentation. It also includes a tutorial and walkthrough.