During its annual Re:Invent conference in Las Vegas, AWS unveiled a number of new services, including a deep-learning based wireless video camera called DeepLens, and a machine-learning based managed service called Amazon SageMaker.
AWS is aiming to make machine learning capabilities accessible to all developers. While DeepLens will enable developers with machine learning experience to run real-time computer vision models, the SageMaker will help developers and data scientists to quickly build, train, deploy, and manage their own machine learning models.
“Our original vision for AWS was to enable any individual in his or her dorm room or garage to have access to the same technology, tools, scale, and cost structure as the largest companies in the world. Our vision for machine learning is no different,” said Swami Sivasubramanian, VP of Machine Learning, AWS. “We want all developers to be able to use machine learning much more expansively and successfully, irrespective of their machine learning skill level. Amazon SageMaker removes a lot of the muck and complexity involved in machine learning to allow developers to easily get started and become competent in building, training, and deploying models.”
DeepLens is a fully programmable and custom-designed HD video camera, powered by Intel Atom X5 Processor which can run more than 100 deep learning operations per second.
It uses Intel’s deep learning software tools and libraries including Intel clDNN (Compute Library for Deep Neural Networks), to run real-time computer vision models directly on the device.
The custom-designed feature of DeepLens allows developers to create their desired deep learning projects with AWS Lambda functions, e.g. it can be programmed to identify numbers of license plates, or text the owner when their dog is on the couch.
Since DeepLens includes example code, sample projects, and pre-trained modes, the developers with no machine learning experience can also run their first deep learning model in just ten minutes.
Amazon SageMaker, the fully managed machine learning model, includes few of the mostly used deep learning algorithms including factorization machines, and linear regression. Developers can use algorithms according to their data source, and get respective drivers and frameworks installed & configured automatically.
Developers can explore and visualize the data stored in Amazon S3 (Amazon Simple Storage Service), and transform it using the libraries, frameworks, and interfaces.
SageMaker enables the developers to select the Amazon EC2 instances with the location specification, making the training easier and faster. It automatically scales EC2 instances across Amazon Availability zones (AZs), performs health checks, applies security patches, and conducts routine maintenance, so that models can be deployed in production with one-click.
AWS has also introduced new speech, language, and vision services for the developers who don’t have expertise in machine learning.
Amazon Transcribe- It converts the speech and audio stored in Amazon S3 into accurate and fully punctuated text. It can also convert the low fidelity audio, and currently supports English and Spanish.
Amazon Translate- It uses neural machine translation techniques to convert text from one language to another. It currently supports English, Spanish, Arabic, French, German, Portuguese, and Chinese, with more to be added in 2018.
Amazon Comprehend- Integrated with deep learning techniques, it can identify text entities, documents, language of the text, sentiments in text, and key phrases with adjectives such as ‘beautiful’, ‘warm’, etc.
It can read data from Amazon S3, Amazon RedShift, Amazon Relational Database Service, Amazon DynamoDB, etc.
Amazon Rekognition Video- It offers real-time recognition of faces for live stream videos (even when they are partially hidden from view), and accurately detect thousands of activities and objects from videos streams and video content stored in Amazon S3.
DeepLens and SageMaker will ship next year. While Translation and Transcribe are available in preview, Comprehend and Rekognition Video are available generally.