Advanced Computing in the Age of AI | Tuesday, July 23, 2024

Amazon, Intel Collaborate on Deep Learning 

Amazon Web Services and Intel have collaborated on a wireless video camera intended to serve as a development tool for new AI and machine learning applications.

The public cloud giant and chipmaker unveiled their programmable "DeepLens" camera during this week's AWS re:Invent conference in Las Vegas. The programmable camera serves as a template for creating what the partners see as a range of "smart" products for industrial, retail and consumer users, including smart home applications.

Intel (NASDAQ: INTC) said DeepLens combines its Atom X5 processor with a user interface that supports the training and deployment of machine learning models in the cloud. Intel promotes its Atom X series processors for low-power mobile applications. The wireless camera also includes embedded graphics to support applications such as object detection and identification, it said.

In addition, the camera uses Intel's deep learning software tools and libraries such as the chipmaker's Math Kernel Library. Those components help run real-time computer vision models on the DeepLens camera.

The tools are intended to accelerate the design of AI and machine learning products using preconfigured frameworks built into DeepLens. Intel said the development platform initially supports the Apache MXNet deep learning library. Tensorflow and Caffe2 support will be added in the first quarter of 2018.

AWS (NASDAQ: AMZN) is an early contributor to Apache MXNet, which is billed as a scalable training and inference framework. The Amazon distribution of MXNet includes an interface to Gluon, the open source deep learning library designed, among other things, to accelerate development of models used to construct neural networks. It includes an API providing access to a library of neural network components used to define machine-learning models.

Last month, AWS and cloud rival Microsoft said they would collaborate on an "Open AI" project based on Gluon. Neural networks based on Gluon could use AWS and Microsoft (NASDAQ: MSFT) distributed training tools. Optimized inference would also allow deep learning models to run on lower end hardware, they added.

Gluon is available as part of Amazon's Apache MXNet deep learning framework; Microsoft is planning to release it as part of its Cognitive Toolkit.

The collaboration with Intel, Microsoft and others is part of a broader push by AWS into deep learning development via its Amazon Machine Images initiative. The images provide a software configuration required to launch an instance on its Elastic Compute Cloud.


About the author: George Leopold

George Leopold has written about science and technology for more than 30 years, focusing on electronics and aerospace technology. He previously served as executive editor of Electronic Engineering Times. Leopold is the author of "Calculated Risk: The Supersonic Life and Times of Gus Grissom" (Purdue University Press, 2016).