Highlights from the AWS annual conference
- By Susan Miller
At its annual re:Invent conference in Las Vegas, Amazon announced a number of services that can help public sector agencies take better advantage of machine learning, robotics and satellite management.
A new program to help independent software vendors achieve a number of federal authorizations, including those from the Federal Risk and Authorization Management Program, Defense Federal Acquisition Regulation Supplement, the Criminal Justice Information Services.
The company said on its blog that the Authority to Operate on AWS program includes training, tools, pre-built Cloud Formation templates and control implementation details and pre-built artifacts, as well as direct guidance from AWS consulting and technology partners, including GitHub, RedHat, Splunk, the Center for Internet Security and Coalfire.
The AWS Ground Station service allows satellite operators to cut their communication costs by paying only for satellite communications their organization needs. The ground station as a service eliminates the need for universities and government agencies to lease ground antennas across the globe to communicate with their satellites. Customers can control satellite communications, downlink and process satellite data, and easily scale operations without managing ground station infrastructure.
AWS is building out its new Ground Station network with an initial pair of hubs with the intent to have 12 in operation by the middle of next year, Chief Evangelist Jeff Barr said in an AWS blog post.
Meanwhile, Lockheed Martin is linking customers of its Verge antenna network to those ground stations to give them access to AWS services like storage, analytics and machine learning.
The service will be available to AWS customers in the government or highly regulated sectors involved in missions such as public safety, military, data collection and Earth observation. This covers AWS’ GovCloud and Secret Regions across nearly every data classification.
For machine learning applications, Amazon announced several new features for Sagemaker, its fully managed service that helps developers build, train and deploy machine learning models.
Amazon SageMaker Ground Truth addresses the time-consuming and expensive challenge of labeling thousands of examples required to train machine learning models. The service learns in real time from a sample of labels assigned by humans and automatically applies labels to much of the remaining data. Amazon SageMaker Ground Truth reduces costs by up to up to 70 percent when compared to human annotation, the company said.
Amazon SageMaker RL addresses the complexity of reinforcement learning, which uses an algorithm achieve a complex goal that learns from right and wrong decisions as it explores its environment, usually in a simulator. Traditionally complex and expensive, reinforcement learning can be used to find the best health care treatments, optimize manufacturing supply chains or solve gaming challenges, the company said. Amazon SageMaker RL adds pre-packaged reinforcement learning toolkits to SageMaker so developers can more easily build, train, and deploy reinforcement learning algorithms.
Amazon SageMaker Neo, meanwhile, allows machine learning models to trained once and run in the cloud or on connected devices at the network edge with twice the performance and no loss in accuracy, AWS said. That means developers no longer need to modify their trained models for different hardware platforms. The company said it will make Neo available as an open source project.
AWS also announced a machine learning university training and certification program with more than 30 self-service, self-paced digital courses for developers, data scientists, data platform engineers and business professionals.
AWS RoboMaker uses cloud services to help developers to develop, test and deploy robotics applications, as well as build intelligent robotics functions. AWS RoboMaker connects the Robot Operating System to AWS' machine learning, monitoring and analytics services so robots can "stream data, navigate, communicate, comprehend and learn," the company said in its overview.
The service also features an integrated development environment, a fully managed simulation service with pre-built virtual 3D worlds such as indoor rooms, retail stores, and race tracks so users can download, modify, and use these worlds in their simulations. The company said it is piloting AWS RoboMaker with 12 universities.
AWS also announced a $399 autonomous toy car -- AWS DeepRacer -- for developers to test some of their own self-driving programs. Users can train reinforcement learning models in an online simulator and then test-drive them on DeepRacer.
Arizona State University announced an initiative that focuses on building smarter communities in the Phoenix metropolitan area by using the cloud computing, artificial intelligence and machine learning to solve community and regional challenges. The ASU Smart City Cloud Innovation Center aims to improve digital experiences for smart city designers, expand technology alternatives to minimize costs, spur economic and workforce development and facilitating sharing of public sector solutions within the region.
Susan Miller is executive editor at GCN.
Over a career spent in tech media, Miller has worked in editorial, print production and online, starting on the copy desk at IDG’s ComputerWorld, moving to print production for Federal Computer Week and later helping launch websites and email newsletter delivery for FCW. After a turn at Virginia’s Center for Innovative Technology, where she worked to promote technology-based economic development, she rejoined what was to become 1105 Media in 2004, eventually managing content and production for all the company's government-focused websites. Miller shifted back to editorial in 2012, when she began working with GCN.
Miller has a BA from West Chester University and an MA in English from the University of Delaware.
Connect with Susan at firstname.lastname@example.org or @sjaymiller.