Chapter 11. The Future of ML on Mobile Applications

Machine learning (ML) requires massive computational power and, hence, requires specialized processors. But if the power of ML can be brought to mobile devices that lack such processing power and also work in offline mode, there will be enormous opportunities and an entire new business category with a whole gamut of innovative useful mobile applications that are very hard to imagine otherwise. The entire way customers and businesses connect with each other would be reshaped.

Mobile devices have become extended organs of human beings these days. It is hard to find anyone without a mobile phone with them always. If a mobile phone is going to be a part of the human being, then, just as the eyes, nose, legs, and so on know what we do daily and have got accustomed to our lifestyle, in a similar manner, mobile phones can alsounderstand the ins and outs of our daily routine and can bring out so many key data points, which we may not have had the...

Key ML mobile applications 

In this section, we will look at some of the most popular mobile applications and understand what they are doing in the field of mobile ML.

Facebook

Facebook has developed an AI platform, Caffe2Go. Through this toolset, Facebook initially wanted to provide enriched AI and AR experiences to users. They are enabling users to process videos and images through on-device ML and perform certain tasks without having to transmit these videos and images to the backend for complex image and video processing. Their style transfer toolkit enables users to take the artistic qualities of one image style, and apply it to other images and videos.

Google Maps

Google has introduced TensorFlow Lite as well as ML Kit that enables users to perform mobile ML in mobile applications. Google Maps from Google is a classic example of ML on mobile.

Snapchat

Snapchat is innovating on complex ML algorithms that are able to perceive facial features on an image captured by the camera. These algorithms...

Key innovation areas

The following sections detail some of the business areas where innovation is happening, leveraging the power of ML. A number of players are already leading the way in this regard.

Personalization applications

Understanding user behavior by leveraging various parameters that are provided through mobile devices and understanding their life patterns for the purposes of personalization will be of value to users. When the same mobile application is going to cater to user profiles across a broad spectrum, it will be of significant value if it could provide specific features that best suit the person using it. Such advanced personalization could be brought into applications by leveraging ML.

Healthcare

Here, there are various use cases that help track various health parameters that can be tracked, learned, and put into use for providing innovations in healthcare, such as diagnostic applications that can diagnose based on pictures and sound from mobile applications.

Fitness tracking...

Opportunities for stakeholders

This section provides details of the key stakeholders in the landscape who contribute and determine the success and spread of ML on mobile devices. It explores how they contribute to mobile ML and what innovations are being carried out by each of them to increase the acceptance of mobile ML and make it reach far and wide.

Hardware manufacturers

The hardware is the platform that forms the basis for executing ML mobile applications. ML has specific requirements in terms of processing units and memory in order to run the complex ML algorithms. Until recently hardware limitations was one reason that drove the majority of ML processing to be undertaken in backend servers where there are no limits on processing units or memory. But now, most device manufacturers are making groundbreaking innovations that render hardware suitable for running mobile on-device ML applications:

• Apple has already designed and built a neural engine as part of its iPhone X's main chip set...

Summary

In this chapter, we learned about the future of ML in the field of mobile and how it will be useful to users. We also discussed different mobile applications that use ML, including Facebook, Netflix, and Google Maps.

We also saw how a variety of business areas are using ML applications and the various opportunities that exist for stakeholders in the field of ML using mobile.