Time to deep learn? From Dreamify to Siri, the applications appear endless


Mike Dano The tools that are available to the world's mobile developers appear to be expanding dramatically, from virtual reality kits to those for the Internet of Things. But a new utensil released in just the past few months -- and bolstered by an online learning course announced in recent days -- could prove far more useful to developers interested in more advanced services and applications.

But let's start at the beginning. From my cursory research into the topic, deep learning appears to use computers to sift through large amounts of data and come to logical, or at least interesting, conclusions. The technology helps to underpin a wide range of services, from Google's photo search to the speech recognition functions of Apple's Siri and Microsoft's Cortana.

Or, as the more tech-minded folks from Wikipedia explained, deep learning is "based on a set of algorithms that attempt to model high-level abstractions in data by using multiple processing layers with complex structures, or otherwise composed of multiple non-linear transformations."

Google has long been playing in this area, and last year it open sourced its TensorFlow software to encourage others to work in the area of machine and deep learning. The technology could well have major implications: "In literally giving the technology away, Google believes it can accelerate the evolution of AI [artificial intelligence]. Through open source, outsiders can help improve on Google's technology and, yes, return these improvements back to Google," Wired said.

For developers, an example of this type of technology in action comes from Dreamify, an app released for iOS and Android last year that uses a similar type of neural networking that TensorFlow does. Dreamify uses Google's DeepDream source code and Amazon web servers to run remotely. The result is an app that can transform users' images into psychedelic versions of the original, like a Salvador Dalí Instagram filter.

Deep learning appears to have a wide range of applications, from speech recognition to image recognition to natural language processing. It may also be used to improve customer relationship management software, recommendation systems and more.

In recent days, Google and Udacity announced a free, three-month online course on deep learning geared toward encouraging developers and others to dip their toes into the technology. Vincent Vanhoucke, a manager in Google's deep learning infrastructure team, said that he got involved in the deep learning scene five years ago through his work in Google Voice Search. "At that time, I didn't use Google's bazillion machines to get started with deep learning: I bought a modest computer with a GPU," he said. "Getting started was difficult then: few people outside of select research groups in academia knew the tricks of the trade which were necessary to make deep learning work well. But the trend -- that continues today -- of researchers using open source tools, and open sourcing the results of their papers started to take root, and today that knowledge is readily accessible to anyone with basic understanding of machine learning."

As the mobile app market matures, and profits become increasingly more elusive, areas like virtual reality, the IoT and, yes, deep learning may well be where adventurous developers might want to allocate their time. --Mike