Gluon is now compatible with Apache MXNet, AWS's preferred machine learning framework, and Microsoft is working to enable its compatibility with its Cognitive Toolkit. With the Gluon interface, developers can build machine learning models using a simple Python API and a range of pre-built, optimized neural network components. That's particularly important for increasingly complex models, since the calculations needed to both train and execute a neural network can take significant amounts of compute power. "We look forward to our collaboration with Microsoft on continuing to evolve the Gluon interface for developers interested in making machine learning easier to use". Conversely, some engines focus on offering simpler model building at the expense of training performance.
Machine learning models are crucial to how artificial intelligence systems work today: essentially developers build these models to help run different services, whether they be messaging bots, scripts for voice-based home hubs, face-recognition apps or autonomous driving systems. Gluon aims to strike a balance between the two, giving developers an easy programming interface and quick training. Developers can use the Gluon interface to create neural networks on the fly, and to change their size and shape dynamically.
Gluon was developed together between the two companies, allowing developers to quickly build machine learning models without compromising training performance.
Because Gluon is backed by Microsoft and AWS's respective deep learning frameworks, it's also possible for developers to distribute neural network computation from Gluon models across many GPUs for increased speed.
"The potential of machine learning can only be realized if it is accessible to all developers", said Swami Sivasubramanian, vice president of Amazon AI, in prepared remarks. "I really appreciate how the Gluon interface is able to keep the code complexity at the same level as the concept; it's a welcome addition to the machine learning community".
Gluon supports symbolic and imperative programming, which is something not supported by many other toolkits, Microsoft explained. First, more deep learning providers will need to add support for Gluon, letting developers decide which AI toolchain to use. Both Amazon and Microsoft said they will continue to collaborate on Gluon, presenting it as a way of creating an "open AI ecosystem" that's accessible to the entire industry.