Alphabet’s Google is about to embrace AI to maintain Android forward of the sport with its Apple counterpart. We get its indication within the company’s latest announcement at Google I/O 2017, an annual developer convention held in Mountain View, California. In an unlimited array of bulletins unfold throughout the vast service spectrum, one assertion about Android was: TensorFlow Lite will permit app builders to entry some superior neural networking options in Android OS. The assertion clearly reveals the tech big’s intentions to go forward with AI (Synthetic Intelligence) and machine studying implementation within the Android apps.

Let’s perceive how Google is about to organize Android for the long run anticipated to be dominated by AI:
Google has already begun investing heftily within the ML (Machine Studying)-specific {hardware} in tune with Cloud Tensor Processing Unit (TPU) chips foreseeing the ‘An I First’ world. Such {hardware} can speed up each the coaching of latest ML algorithms and information processing whereas utilizing present fashions. The corporate has introduced the improved second model, TensorFlow Lite, which may run on smartphones and different cellular gadgets, within the latest Google I/O 2017 convention.

How the Lite model will pave method for AI in Android

Dave Burke, VP of Engineering for Android is of the opinion that the Lite model of TensorFlow will leverage a neural community API to faucet into silicon-specific accelerators. In his phrases: “We expect these new capabilities will assist energy the following technology of on-device speech processing, visible search, augmented reality, and extra.” He outlined TensorFlow Lite as a library for Android apps designed to make quick and small apps with state-of-the-art strategies.
Google needs to make the Lite framework part of its open supply mission TensorFlow and to launch the neural community API as the following main Android replace within the later a part of 2017. It was the final yr when Google enriched TensorFlow with Android and iOS help, this yr, the corporate has taken the following step with TensorFlow Lite.

The way it works

AI-focused chips within the Lite model could make it attainable for Android telephones to deal with superior machine studying computations with out utilizing a lot energy. Apparently, as increasingly Android apps will use machine studying for providing clever experiences to the customers, this type of work will change into simpler on the system.
Proper now, it’s troublesome to combine superior machine studying into apps, particularly coaching fashions, as the method requires extra computational energy and a number of time. Sturdy {hardware} and excessive programming abilities are obligatory for integrating ML within the purposes. As a consequence of these causes, virtually it was not attainable to incorporate machine studying in shopper cellular purposes. Such apps typically offload the method to an enormous information heart. Now, information course of within the cloud has a number of downsides as properly: customers ought to be prepared to switch their information like photographs, textual content, and so forth. to an organization’s servers and require wealthy connectivity to make sure the low latency of course of.

The Lite model of TensorFlow is designed to supply an answer to all these difficulties as a machine learning-specific Digital Sign Processor (DSP). Utilizing this model, apps will get smarter and do attention-grabbing issues extra rapidly on their very own with out relying on information facilities. What’s extra, as clever apps, they’ll refine picture or speech-recognition on the transfer, and Android gadgets could have extra computing energy.

We are able to actually anticipate extra chips, specializing in the good gadgets, for accelerating machine studying coaching and inference associated to synthetic intelligence.

Wrapping up

In at the moment’s mobile-driven world, Google can get the advantages from integrating AI techniques into Android apps as Apple is way from integrating such AI-friendly instruments into iOS purposes. It’s now a matter of time when Google will convey a radical change in the appear and feel of Android apps.