Introduction to Machine Learning on Mobile

We're living in a world of mobile applications. They've become such a part and parcel of our everyday lives that we rarely look into the numbers behind them. (These include the revenue they make, the actual market size of the business, and the quantitative figures that would fuel the growth of mobile applications.) Let's take a peek at the numbers:

• Forbes predicts that mobile application revenue is slated to hit $189 billion by the year 2020
• We are also seeing that the global smartphone installation base is increasing exponentially. Therefore, the revenue from applications getting installed on them is also increasing at an unimaginable rate

Mobile devices and services are now the hubs for people's entertainment and business lives, as well as for communication. The smartphone has replaced the PC as the most important smart connected device. Mobile innovations, new business models, and mobile technologies are transforming every walk of human life.

Now, we come to machine learning. Why has machine learning been booming recently? Machine learning is not a new subject. It existed over 10-20 years ago, so why is it in focus now and why is everyone talking about it? The reason is simple: data explosion. Social networking and mobile devices have enabled the generation of user data like never before. Ten years ago, you didn't have images uploaded to the cloud like you do today because mobile phone penetration then cannot be compared to what it is today. The 4G connection makes it possible even to live stream video data on-demand (VDO) now, so it means more data is running all around the world like never before. The next era is predicted to be the era of the internet of things (IOT), where there is going to be more data-sensor-based data.

All this data is valuable only when we can put it to proper use, derive insights that bring value to us, and bring about unseen data patterns that provide new business opportunities. So, for this to happen, machine learning is the right tool to unlock the stored value in these piles and piles of data that are being accumulated each day.

So, it has become obvious that it is a great time to be a mobile application developer and a great time to be a machine learning data scientist. But how cool would it be if we were able to bring the power of machine learning to mobile devices and develop really cool mobile applications that leverage the power of machine learning? That's what we are trying to do through this book: give insights to mobile application developers on the basics of machine learning, expose them to various machine learning algorithms and mobile machine learning SDKs/tools, and go over developing mobile machine learning applications using these SDKs/tools.

Machine learning in the mobile space is a key innovation area that must be properly understood by mobile developers as it is transforming the way users can visualize and utilize mobile applications. So, how can machine learning transform mobile applications and convert them into applications that are any user's dream? Let me give you some examples to give a bird's eye view of what machine learning can do for mobile applications:

• Facebook and YouTube mobile applications use machine learning—Recommendations or People you might know are nothing but machine learning in action.
• Apple and Google read the behavior or wording of each user behavior and recommend the next word that is suitable for your style of typing. They have already implemented this in both iOS and Android devices.
• Oval Money analyzes a user's previous transactions and offers them different ways to avoid extra spending.
• Google Maps is using machine learning to make your life easier.
• Django uses machine learning to solve the problem to find a perfect emoji. It is a floating assistant that can be integrated into different messengers.

Machine learning can be applied to mobile applications belonging to any domain—healthcare, finance, games, communication, or anything under the sun. So, let's understand what machine learning is all about. 

In this chapter, we will cover the following topics:

• What is machine learning?
• When is it appropriate to go for solutions that get implemented using machine learning?

• Categories of machine learning
• Key algorithms in machine learning
• The process that needs to be followed for implementing machine learning
• Some of the key concepts of machine learning that are good to know
• Challenges in implementing machine learning
• Why use machine learning in mobile applications?
• Ways to implement machine learning in mobile applications