An aspect of artificial intelligence (AI) research, machine learning is focused on the process by which computers improve at performing tasks by exposure to data rather than by specific programming. Algorithms are constructed to allow the machine to collect and analyze data, comparing large numbers of examples to discover patterns. This allows the machine to draw conclusions and learn from its accumulated experience in order to predict results in new data. The result? “Instead of people writing software, we have data writing software,” says Jen-Hsun Huang, Chief Executive of NVIDIA, a firm that makes chips for AI systems.
This hasn’t only led to change in research and software development, but also to the daily lives of people across the planet. Read on to discover the greatest implications of machine learning for mobile applications.
AI is already a part of our day-to-day experience. Over the past decade or two, consumers have become accustomed to seeing the results of machine learning in the form of recommendations from online retailers, or music streaming and video services based on the user’s previous searches, purchases, and ratings. Machine learning is used for many other commonplace applications such as search engines, identifying and blocking spam email, detecting text in a foreign language and providing a translation, or recognizing unusual activity to detect potential credit card fraud. The progress made in this area provides a foundation for advanced uses of the technology, from revolutionary self-driving cars and drones to medical diagnostics, legal research, and financial advising.
In December 2016, Bank of America Merrill Lynch published Robot Revolution – Global Robot & AI Primer with the goal of “setting out the challenges and opportunities offered by robots and AI.” The paper outlines the potential economic effects that may result from the increasing use of robots and AI, as well as possible investment opportunities. It demonstrates the massive scale of AI’s influence and projected growth over the next several years. They estimate that by 2020, the AI and robotics market will expand to $153 billion, up from $58 billion in 2014.
The Internet of Things, which includes physical devices, vehicles, buildings and other items which are connected to and can exchange information over the internet, will play a big part in this. While some of this market consists of industrial uses for AI and robotics, a significant portion of the market is expected to be made up of business, home, and personal uses like mobile applications.
A Pocket Full of AI
Neural networks are designed to mimic the structure and function of neurons and their connections in the human brain. They are used in many familiar applications that rely on machine learning such as voice recognition software, Facebook’s facial recognition software for photo tagging, and Snapchat lenses, to name but a few. Until recently, machine learning relied on massive computing power, which required a connection to a server with enough energy to handle the load. Mobile applications have been limited by this need to use cloud-based servers in order to function. However, new technology is beginning to change this.
In 2016, researchers at MIT announced that they had designed a new chip to implement neural networks that is 10 times more efficient than existing mobile processors. This will allow mobile devices to run AI algorithms internally without connecting to the internet. The offline process saves time and energy and means much more powerful apps can be created to use on mobile devices. Speech recognition and language translation applications are already being developed, with other types of apps sure to follow.
Once machine learning applications no longer have to connect to the internet to function, the variety and utility of available apps is likely to expand rapidly. An examination by Deloitte University Press concluded that machine learning applications will soon be found in every industry, and will feature one or more of the following capabilities:
- An analysis or diagnosis of sensory data
- Perceptual interfaces or interactivity
- Navigation and motion control
These new apps could include personal health monitoring, augmented and virtual reality, drone piloting, presentation capabilities, and more. The ability to provide personalized information to an individual while navigating a city street, retail store, airport or museum will surely be of interest to advertisers or the curators of such spaces.
Progress in machine learning is made thick and fast, so it’s never too early to start considering new applications that could enhance your life and business.