This site may earn affiliate commissions from the links on this page. Terms of use.

As AI and deep learning accept gone mainstream, a wide range of companies have announced they'll bring compatible products to market. Anybody from Google and Nvidia to AMD and Fujitsu have thrown their hats into this particular ring. But the software that runs on deep learning and AI-specific hardware is nonetheless typically a custom solution adult by individual companies. Microsoft and Facebook are teaming up to change that, with a new common framework for developing deep learning models.

The Open Neural Network Exchange (ONNX) is described as a standard that will let developers to motility their neural networks from 1 framework to another, provided both adhere to the ONNX standard. According to the joint printing release from the ii companies, this isn't currently the instance. Companies must choose the framework they're going to use for their model earlier they start developing it, but the framework that offers the all-time options for testing and tweaking a neural network aren't necessarily the frameworks with the features y'all desire when you bring a product to market place. The press release states that Caffe2, PyTorch, and Microsoft'south Cognitive Toolkit will all back up the ONNX standard when information technology'south released this month. Models trained with 1 framework will be able to move to another for inference.

Facebook's side of the post has a bit more item on how this benefits developers and what kind of code compatibility was required to support information technology. It describes PyTorch as having been built to "push the limits of enquiry frameworks, to unlock researchers from the constraints of a platform and let them to express their ideas easier than before." Caffe2, in dissimilarity, emphasizes "products, mobile, and extreme functioning in listen. The internals of Caffe2 are flexible and highly optimized, and so we can transport bigger and meliorate models into underpowered hardware using every trick in the volume." By creating a standard that allows models to move from one framework to another, developers are able to take advantage of the strengths of both.

There are still some limitations on ONNX. It'southward non currently compatible with dynamic menses control in PyTorch, and FB alludes to other incompatibilities with "advanced programs" in PyTorch that it doesn't particular. Still, this early endeavor to create common basis is a positive step. Most of the ubiquitous ecosystems we take for granted — USB compatibility, 4G LTE networks, and Wi-Fi, just to proper noun a few — are fundamentally enabled by standards. A silo'd go-it-solitary solution can work for a company developing a solution it but intends to employ internally, merely if you want to offer a platform others can use to build content, standardizing that model is how you encourage others to use it.

The major difference between Microsoft and the other companies developing AI and deep learning products is the difficulty Redmond faces in baking them into its consumer-facing lineup. With Windows 10 Mobile finer dead, MS has to rely on its Windows marketplace to drive people towards Cortana. That'south an intrinsically weaker position than Apple or Google, both of which have huge mobile platforms or Facebook, which has over a billion users. ONNX should benefit all the players working on AI, but it's probably more of import for MS than for other companies with larger user bases. When you own the most popular phone OS on World, you don't have to worry much about whether someone else's neural network models play nicely with yours.

At present read: Artificial neural networks are irresolute the earth. What are they?