In a world where Google’s AlphaGo AI takes a human down in the most sophisticated open board game, Stephen Hawking warns that machines could end mankind, and Elon Musk is promising free artificial intelligence for all, it goes without saying that machine learning is a hot trend.
Today we are much closer to the creation of true artificial intelligence than ever before, but the edge of the AI frontier still looms somewhere in an uncertain future. However, one element of artificial intelligence is already improving mobile and web services today: machine learning.
In this article, we’ll discuss how machine learning has developed in recent years, and will get a glimpse at where it is headed.
What is machine learning? (in short)
While the ultimate goal of artificial intelligence (AI) studies is the creation of a machine that can mimic a human mind, the idea behind machine learning is constructing computer algorithms that automatically improve themselves by finding patterns in existing data without explicit instructions. In other words, machine learning is more akin to data mining and statistical analysis than it is to classic “AI.”
Machine learning relies entirely on data. The more data (and the higher the quality of data) an algorithm has, the more accurate it becomes. Machine learning becomes more and more effective as the quantity of data in our world grows over time. Our technological capabilities also have to increase in order to keep up with that growth.
Machine learning has been around for a long time. In 1952, Arthur Samuel wrote the first computer learning program that could play checkers. In the 1990’s, scientists turned from a knowledge-driven approach to a data-driven approach, developing machine learning as we know it today. Since the 90’s, the use cases for machine learning technologies have grown year over year. As the field has expanded, a new machine learning subfield — deep learning — has emerged. Deep learning deals with object recognition in images and video.
Today, with rapid technology advances, machine learning has found practical applications in the fields of health care, robotics, surveillance, and, of course, mobile and web apps.
The applications of machine learning in modern digital industry
Many technology companies — both large and small — are now using machine learning tools for image and speech recognition, natural language processing, spam detection and filtering, and advertising.
User personalization is currently the main commercial application of machine learning in the civilian tech sector. Personalization takes on a wide variety of forms, but the main idea is always to give users some sort of content intended specifically for them. This personalization is accomplished by processing massive quantities of user data: personal information, search engine history, content interactions, and all kinds of other information.
“People you might know” on social networks, merchandise you “might be interested in” on an eCommerce site, targeted advertisements, and suggested posts in your Facebook feed — all this content is determined by machine learning algorithms.
As an example of how such personalization can work, here’s our extensive blog post about how machine learning can be utilized in creating a style advice app.
Machine learning algorithms are able to detect and recognize objects depicted in images, including human faces. Object detection is commonly used today for image search. Facial recognition has a variety of applications within chat apps, photo editing apps, dating apps, user authentication, real-time translation and so on. In 2015, Facebook launched a feature that uses image recognition algorithms to describe images to those who are vision-impaired.
Apple’s Siri, Google’s Google Now, Facebook’s M, Microsoft’s Cortana and Amazon’s Alexa all use machine learning and voice recognition algorithms to provide their assistant services to users. Considering the wild competition, tech giants will likely invest significant resources into making their digital assistants even more powerful in the near future.
Predictive analytics to increase app engagement and monetization
A distinct branch of machine learning, predictive analytics is the practice of using advanced statistics and historical data to predict future outcomes. There are a number of mobile analytics services that use this approach, including Amazon Mobile Analytics and Amazon Machine Learning, Microsoft Azure, Google Cloud Platform Machine Learning, and Agilone.
Analyzing user interactions within an app, predictive analytics provides you with a better understanding of which users are likely to remain engaged, which may churn, and which are the most likely to convert.
By understanding market patterns an organization can identify profitable opportunities, offer their customers the products and content they want, and avoid risks in the process.
What comes in the future
Here are only few applications of machine learning that could improve the quality of mobile services in the near future.
We already mentioned that personalization is the most salient application of machine learning today — but there’s still plenty of room for improvement. In the future, users will receive more precise recommendations and ads will become both more effective and less annoying.
Neural networks running on our mobile devices
In 2016, MIT researchers presented an energy-efficient chip that’s able to implement functional neural networks on smartphones and other mobile devices. This means that a future mobile device may have the ability to conduct machine learning tasks locally, opening up a wide range of opportunities for object recognition, speech, face detection, and other innovations for mobile platforms.
Meanwhile, Google and tech company Movidius have announced next-gen devices with machine learning and deep learning capabilities with the help of vision processing units (VPU).
Vision processing lets machines view, interpret and “understand images,” including understanding an image’s symbolism and context. It’s still unknown how these chips can be utilized in mobile devices. At the very least, they always process some fancy images.
Mobile experience automation
There are a lot of apps that automate the work of different connected apps (like IFTTT) or the device’s OS the whole (like Tasker). However, such apps may be clumsy or difficult to use. What if a device could be automated by machine learning algorithms? And what if this automation could be extended to the Internet of Things?
Google already patented a similar idea back in 2012, so it’s possible we’ll see an implementation of this sort of technology sooner or later.
Real-time speech translation
In late 2014 Skype launched Skype Translator. It’s been improving the service since then, and currently provides real-time audio translation among seven languages. If this technology continues to develop, it could significantly improve the quality of international communication or even eradicate language barriers.
Health and fitness
Fitness tracking wearables and apps are pretty popular right now. People gladly use wearables and connected apps to track their sport activities and everyday life. Machine learning has the potential to take this a step further, however, by providing more detailed feedback and tips about a user’s activity and condition, making fitness trackers more effective.
Prolonging a mobile device’s battery life
This may sound a lot less epic than other possibilities of machine learning, but preserving battery life is one of the most frustrating concerns for mobile app users. Along with the automation of system resource allocation for apps, machine learning could also reduce the amount of unnecessary battery consumption by apps.
Current machine learning technologies are already quite powerful, and integration of machine learning into our everyday life will only deepen over the next few years. Quite likely, this will increase our dependence on technology even more, and also raise lots of privacy and ethical questions since there will always be those who try to use machine learning for malevolent purposes. But if used responsibly and focused on providing better services and better mobile experiences, machine learning will mutually benefit both entrepreneurs and mobile users.