LAS VEGAS — Amazon Web Services released a tool this week to empower developers to build smarter, artificial intelligence-driven applications like the AI experts.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
Among the deluge of technologies introduced here at AWS re:Invent 2017, the company’s annual customer and partner event, is a tool called SageMaker. Its function is to help developers add machine learning services to applications.
Machine learning is an artificial intelligence technology that enables applications to learn without being explicitly programmed, and become smarter based on the frequency and volume of new data they ingest and analyze. Few developers are experts in machine learning, however.
SageMaker is geared to that audience. It’s a fully managed service for developers and data scientists who wish to build, train and manage their own machine learning models. Developers can choose among ten of the most common deep learning algorithms, specify their data source, and the tool installs and configures the underlying drivers and frameworks. It natively integrates with machine language frameworks such as TensorFlow and Apache MXNet and will support other frameworks as well.
Alternatively, developers can specify their own algorithm and framework.
The National Football League said it will use SageMaker to extend its next-generation stats initiative to add visualizations, stats and experiences for fans, as well as provide up-to-date information about players on the field, said Michelle McKenna-Doyle, the NFL’s senior vice president and CIO, here this week.
To supplement SageMaker, AWS created DeepLens, a wireless, deep-learning-enabled, programmable video camera for developers to hone their skills with machine learning. One example of DeepLens cited by AWS included recognizing the numbers on a license plate to trigger a home automation system and open a garage door.
AWS’ goal is to democratize access to machine learning technology for developers anywhere, so that individual developers could have access to the same technology as large enterprises, said Swami Sivasubramanian, vice president of machine learning at AWS.
SageMaker is one example of this, said Mark Nunnikhoven, vice president of cloud research at Dallas-based Trend Micro.
“I’ve worked with those technology stacks quite a lot over the last decade and there’s so much complexity …, but now any user doesn’t have to care about it,” he said. “They can do really advanced machine learning very, very easily.”
AWS ups the ante for AI
The general pattern in the market for AI application development has been twofold, said Rob Koplowitz, an analyst at Forrester Research in Cambridge, Mass. There are AI frameworks for data scientists that are extremely flexible but require special skills, and higher-level APIs that are accessible to programmers — and in some cases even non-programmers.
“Amazon wants to provide a middle ground with more flexibility,” Koplowitz said. “It’s an interesting approach and we’re looking forward to getting real work feedback from developers.”
AWS has to play catch-up here with other cloud platform companies that wish to bring machine learning to mainstream programmers. IBM provides developers access to its Watson AI services, and Microsoft has its Cognitive Services and Azure Machine Learning Workbench tools. Reducing the complexity of building machine learning models is among the more difficult areas for businesses, so this is a step in the right direction for AWS, said Judith Hurwitz, founder and CEO at Hurwitz & Associates in Needham, Mass.
Computational intelligence in general, and AI and deep learning in particular, is a hot market with a small community of experts among the biggest tech companies from Facebook to IBM.
“They all have a lot of the same core competencies, but they’re distributing them in different ways,” said Trend Micro’s Nunnikhoven.
Google tends to be more technical, while AWS now wants to make AI more accessible. Microsoft targets specific business analytics uses for AI, IBM wants to show more real-world use cases in areas such as healthcare and financial services, and Apple is looking at AI for privacy and devices. But they’re all contributing back to the same projects, such as Apache Mahout and Spark MLlib, Google’s TensorFlow, Microsoft’s Cognitive Toolkit, and others.
SageMaker should help alleviate developers’ fears that data scientists will make them into second-class citizens, but AWS may have aimed too low with SageMaker, said Holger Mueller, principal analyst at Constellation Research in San Francisco. He said he believes it’s more of a kit to empower business users to create machine learning applications.
Other AWS AI-based services
Other AI-enabled AWS services unveiled this week include Amazon Comprehend, a managed natural language processing service for documents or other textual data that integrates with other AWS services to provide analytics, and Amazon Rekognition Video, which can track people and recognize faces and objects in videos stored in Amazon S3.
There are two services now in preview — Amazon Transcribe, which lets developers turn audio files into punctuated text, and Amazon Translate, which uses neural machine translation techniques to translate text from one language to another. Translate currently supports English and six other languages — Arabic, French, German, Portuguese, Simplified Chinese and Spanish — with more languages to come in 2018.