Amazon's DSSTNE deep learning software now open source

open source keyboard button

Amazon has decided to follow in the footsteps of Google and other technology companies by open-sourcing its deep learning software.

The company has released its deep learning software DSSTNE (pronounced destiny) on GitHub under an open-source Apache license. Deep learning has gained a lot of traction in recent months and many tech companies are currently developing their own software to help teach computers.

Around five months ago, Google made its own deep learning library TensorFlow open-source as well. Amazon included a frequently asked questions page along with its library to explain why it has decided to open-source DSSTNE: "We are releasing DSSTNE as open source software so that the promise of deep learning can extend beyond speech and language understanding and object recognition to other areas such as search and recommendations. We hope that researchers around the world can collaborate to improve it. But more importantly, we hope that it spurs innovation in many more areas".

Amazon’s deep learning software does have its limitations and currently it is unable to support convolutional workloads for image recognition and has limited support for recurrent neural networks. However, the software is able to utilize two graphics processing units (GPUs) simultaneously which other frameworks are unable to do.

DSSTNE has even show some performance advantages over Google’s TensorFlow with Amazon saying it is able to provide a 2.1X speedup over its competitors software on a g2.8xlarge GPU instance in the public cloud of Amazon Web Services (AWS).

Deep learning and machine learning are the direction the industry is moving in and with Amazon, Google and other companies making their software available through open-source licenses, these technologies will develop at an even faster pace.

Published under license from ITProPortal.com, a Net Communities Ltd Publication. All rights reserved.

Photo Credit: rvlsoft/Shutterstock

Comments are closed.

© 1998-2024 BetaNews, Inc. All Rights Reserved. Privacy Policy - Cookie Policy.