Home » Mobile » Machine Learning is Coming to Your Smartphones
robonaut-machines-dexterous-humanoid-39644

Machine Learning is Coming to Your Smartphones

With Apple’s Siri, Google’s Assistant, and the many other alternatives becoming available for consumers the world over, one could argue that AI and machine learning has already come to our smartphones. However, the current software models only contain a very limited portion of what is possible with machine learning and artificial intelligence with mobile applications.

One of the biggest contributors to the current influx of new machine learning applications has to do with the hardware catching up to what is currently possible with the software.

For years developers and engineers have been theorizing and speculating as to how intelligent learning could and should be programmed, and time and time again have stopped short once realizing that the processors powering computers and smartphones have been inadequate for any true AI or ML concepts.

This is slowly coming to an end, where we see phones such as Google’s Pixel being shipped with Snapdragon 820’s capable of the same functions as an Xbox 360 for instance.

Following Moore’s law of exponential growth, we’re at a tipping point in terms of hardware specifications becoming powerful enough to warrant serious research into machine learning on a consumer level.

Having your smartphone recognize what apps you use most frequently and sort them accordingly barely qualifies as being true machine learning, but we’re getting there.

Below are three great examples of machine learning already in production use, and we’ll explore each one in detail.

Microsoft Bing’s web spam filters

While not the most popular search engine, Microsoft has been developing some really interesting machine learning algorithms, and have been one of the pioneers in this area.

With development and research into this area having been in place for a number of years already, Bing has seen their search ranking algorithms improved by machine learning in terms of identifying spammy domains, parked domains and other low quality websites.

All of these negative websites out there are quite easy for the human to recognize, but having a computer recognize this on it’s own, and learn and improve from it’s decisions pose different challenges.

Also, while it’s possible to develop a simple service that looks for 404 errors and commonly used words on parked domains, it’s not entirely fool proof, and will be relatively easy to bypass.

Having a machine learn and adjust it’s filters based on it’s own performance and results can not just cut in manual moderation costs, but categorize and penalize bad quality sites much faster than we humans ever could.

Facebook’s People You Might Know 

The popular social platform have been utilizing machine learning in terms of suggesting friends and family that you might know.

While the service uses simple functions, such as looking in your e-mail contacts, and those of your friends, there are more advanced techniques in place as well.

For Facebook user’s having enabled location tracking, this has become one of the most accurate ways of determining outside of certainties.

By tracking users who follow similar movement patterns, log on to Facebook from similar IP adresses or locations, and by cross referencing this data with the enormous set of data already having been collected, Facebook has become experts in recognizing potential friends and relatives.

Speech to Text Recognition

Microsoft’s Cortana, Amazon’s Alexa, Apple’s Siri and now lately Google’s Assistant, have been making headlines ever since they were announced.

Although the first versions of Siri were rather simple compared to the current version, it was groundbreaking in the fact that it could quite accurately determine the correct words, and improved with each new word it heard.

Since then Microsoft and Amazon has joined the mix, offering slightly different approaches to the same challenge; how to have machines accurately recognize voices and words, extracting meaning from them.

With Google having entered the fray of machine learning with their Assistant app, consumers are poised to teach the machines how to better learn and recognize voice on a whole, given that there are many more Android users out there than for iOS.

Coupled with the fact that Cortana on desktop PC’s have seen little use when compared to tablet and smartphone usage, smartphones appear to be the lower entry-level of the two systems, suggesting that smartphones and tablets are where we will see the largest progress in the years to come.

The future looks bright

With all the largest IT organisations dedicating resources and money into researching and developing artificial intelligence and machine learning algorithms, we expect great innovation to occur over the coming months and years.

Microsoft has recently announced a new 5000-man strong department focusing solely on Artificial Intelligence, and all that encompasses.

This is an unprecedented large number of resources put towards this single area of focus, and shows how interested Microsoft is in not just developing machine learning and AI that will fuel the next tech boom, but also the financial gains bound to be associated with becoming one of the leaders in this business.

Keeping the future in mind, another exciting announcement recently named Microsoft in a group of high-profile names including IBM, Google, Facebook and Amazon joining forces in a collaboration dedicated to advancing the understanding of the technology.

This partnership will focus on certain aspects, such as coming up with standardization measures allowing for developers and programmers the world over to write code and software that runs on all types of devices in a shared language, rather than each company having their own programming language and structure.

While we suspect that each of the companies will keep their most key research internal, it makes sense to work together in creating and developing a universal system that everyone can use, instead of the mix and match world of websites and mobile application development we know today.

Just think of how Android apps widely differ from iOS apps, and how a website can look great in Google Chrome, but terrible in Firefox, and you’ll most likely agree that the Partnership on Artificial Intelligence to Benefit People and Society is a good thing, even though the name of the alliance doesn’t exactly roll off the tongue.

Machine Learning is Coming to Your Smartphones b4c5527322c7410c9f1fc88f573ecd36 s 80 d http 3A 2F 2Fwww
Follow Mark

Mark Pedersen

App Developer at Nodes
Mark has been developing web and mobile applications since the early 2000's, always with a penchant for open-source technologies. These days he primarily works with Android apps.
Machine Learning is Coming to Your Smartphones b4c5527322c7410c9f1fc88f573ecd36 s 80 d http 3A 2F 2Fwww
Follow Mark

Latest posts by Mark Pedersen (see all)

Check Also

Mobile App vs. Mobile Website: What’s Your Call

The mobile market is growing bigger and better with every passing year. In 2016, businesses …

Leave a Reply

Your email address will not be published. Required fields are marked *

Time limit is exhausted. Please reload CAPTCHA.

Don't miss the latest news from GEEKERS Magazine.
Signup today for free and be the first to get the latest news.
  Thank you for staying with us.