News
AWS Opens More AI Capabilities to App Developers
At its re:Invent conference this week, Amazon Web Services (AWS) described new services aimed at helping developers add more artificial intelligence (AI) functionality to their apps without extensive coding or machine learning expertise.
The company announced four new AI offerings at various stages of availability: Amazon Personalize, Amazon Forecast, Amazon Textract and Amazon Comprehend Medical.
Amazon Personalize, now in preview, is a recommendation service that's adapted from the same technology that supplies product recommendations to Amazon.com e-commerce customers. Personalize aims to help developers of all experience levels create and deploy recommendation models to fit a wide variety of use cases, including news articles, retail products and digital media.
"Using AutoML, a new process that automates complex machine learning tasks, Personalize performs and accelerates the difficult work required to design, train, and deploy a machine learning model," wrote Julien Simon, an AI and machine learning evangelist for AWS, in a blog post.
Personalize works with datasets that are streamed in real-time, as well as data stored in a customer's Amazon S3 bucket. It comes with built-in recommendation algorithms (which Simon called "recipes") designed for common scenarios, or users can create their own.
Also in preview is Amazon Forecast. Like Personalize, Forecast is built on existing technology used in Amazon.com's e-commerce business. Forecast provides "time-series forecasting," meaning it uses historical trends and time-dependent data to predict future outcomes.
Forecast is a managed service, eliminating the need for users to build their own forecasting models, provision the required compute resources, or pay upfront for resources they may not use.
"Amazon Forecast requires no machine learning experience to get started. You only need to provide historical data, plus any additional data that you believe may impact your forecasts," AWS said in its product page. "Once you provide your data, Amazon Forecast will automatically examine it, identify what is meaningful, and produce a forecasting model capable of making predictions that are up to 50% more accurate than looking at time series data alone."
A new natural language capability that's also in preview is Amazon Textract, which scans documents of "virtually any type" and pulls out relevant data using machine learning. Textract promises to enable users to process "millions of document pages in hours," without any manual coding.
Amazon Comprehend Medical, meanwhile, is a Textract-like service, but for health care. Amazon Comprehend Medical can scan unstructured medical documents like patient records, health providers' notes and clinical trials. Using machine learning, it then extracts relevant medical information from those documents, such as "medications, medical conditions, test, treatment and procedures (TTP), anatomy, and protected health information (PHI)," according to AWS' product page.
Amazon Comprehend Medical builds on the existing Amazon Comprehend text-analysis service, which was introduced at last year's re:Invent conference. "Amazon Comprehend can understand general-purpose text. However, given the very specific nature of clinical documents, healthcare customers have asked us to build them a version of Amazon Comprehend tailored to their unique needs," Simon explained in another blog post.
Unlike the other three services, Amazon Comprehend Medical is now generally available out of AWS' regions in Northern Virginia, Ohio, Oregon and Ireland.