Monday, September 04, 2017

Access To Machine Learning Tools Through Your Browser

Google has announced that we will be all be able to have access to (and to interact with) a powerful machine learning tools library via its browser, without the need for adding special coding, installations or a backend.

Part of The PAIR Initiative

As a result of Google’s PAIR initiative to study and redesign human interactions with Machine Learning (ML), Google has announced that its open source WebGL-accelerated JavaScript library for ML called deeplearn.js 0.1.0 is being made available through a simple browser.

Why?

Google’s stated reasons for the introduction of browser-based access to ML tools are that it will promote general visibility of and engagement with ML, it will give developers an easy access route to powerful new tools, and it will help drive collaboration on ML learning projects and challenges with the open source community. This could, of course, benefit both developers and Google.

DeepMind

Back in March this year, Google researchers announced that its AI company DeepMind had developed a program that could learn one task after another, using skills it acquired on the way. The ability of DeepMind AI to mirror the learning brain, working out which connections in its neural network had been the most important for the tasks it has learned so far, before moving onto another task, were an important step forward in ML.

The fact that Google is offering everyone the chance to access tools that harness this power through a simple browser represents a new and exciting opportunity.

What Can It Do?

ML is a flexible tool that can be applied to all manner of problems. Examples of some areas where it has been used include interactive explanations, rapid prototyping and visualization, offline computation, object recognition, and language translation.

Demos

The Deeplearn.js homepage https://pair-code.github.io/deeplearnjs/ shows a number of demos (on supported devices) to illustrate the kind of thing that deeplearn.js can do. These include enabling you to use your webcam to play with an image classifier that uses your webcam in real-time and watch the network’s internal representations of what it sees. You can also generate abstract art videos at a very smooth 60 frames per second.

What Does This Mean For Your Business?

This move by Google means that you and your business now have easy access (just through a browser) to some of the very latest ML tools. In the past, access to these kinds of tools and resources was restricted only to those with the (often expensive and complicated) computational resources, and the technical expertise to work with commonly available ML libraries. This initiative by Google is therefore offering businesses a real opportunity to learn about and use ML in a way that could solve long-standing business problems, and create the kind of innovations that could lead to new business opportunities, greater competitiveness, and competitive advantages.

While it is likely to be more complicated than it sounds to implement, the point is that the business benefits to Machine Learning are now within the grasp of SME's, in a similar way to how Google Analytics gave them hitherto unimagined insights into their Web marketing a decade ago.

No comments: