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

Cortana-Feature

Every bit AI and deep learning have gone mainstream, a wide range of companies have announced they'll bring compatible products to market. Everyone 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 developed 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 allow developers to movement their neural networks from one framework to another, provided both attach to the ONNX standard. According to the joint press release from the two companies, this isn't currently the case. Companies must cull the framework they're going to use for their model before they start developing it, but the framework that offers the best options for testing and tweaking a neural network aren't necessarily the frameworks with the features you want when you bring a product to market. The press release states that Caffe2, PyTorch, and Microsoft'south Cerebral Toolkit will all support the ONNX standard when information technology'southward released this month. Models trained with one framework will be able to move to another for inference.

Facebook'due south side of the post has a bit more detail on how this benefits developers and what kind of lawmaking compatibility was required to support it. Information technology describes PyTorch as having been built to "push the limits of research frameworks, to unlock researchers from the constraints of a platform and permit them to express their ideas easier than before." Caffe2, in contrast, emphasizes "products, mobile, and farthermost performance in mind. The internals of Caffe2 are flexible and highly optimized, so we can ship bigger and better models into underpowered hardware using every fob in the volume." By creating a standard that allows models to move from ane framework to another, developers are able to take advantage of the strengths of both.

In that location are all the same some limitations on ONNX. Information technology'south not currently compatible with dynamic menses control in PyTorch, and FB alludes to other incompatibilities with "avant-garde programs" in PyTorch that information technology doesn't detail. Notwithstanding, this early effort to create common ground is a positive step. Virtually of the ubiquitous ecosystems nosotros 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 get-it-alone solution tin can piece of work for a company developing a solution it merely intends to apply internally, but if you want to offering a platform others can use to build content, standardizing that model is how you encourage others to use it.

The major divergence 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 market to drive people towards Cortana. That's 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, just it's probably more of import for MS than for other companies with larger user bases. When you own the nearly popular phone Os on Earth, you lot don't have to worry much virtually whether someone else's neural network models play nicely with yours.

Now read: Artificial neural networks are changing the earth. What are they?