The machine learning as a service market, valued at $613.4 million in 2016, is growing at a compound annual growth rate of 43.7 percent and will be worth $3.755 trillion by 2021. The growing demand for machine learning is being fed by innovations that make the technology increasingly practical for both consumers and businesses. Here are three innovations that are helping make machine learning one of today’s most important technology trends.
Empowering Smartphones with Artificial Intelligence
One significant trend is the coming of artificial intelligence (AI) to mobile devices. Until recently, the amount of data processing power required for AI applications depended upon either using a large amount of local server space or tapping into remote cloud servers. Bringing AI to smartphones requires significantly increasing the native computing capability of mobile devices.
To facilitate this, smartphone component manufacturers have developed processors designed to handle the large workload required by AI. For instance, the machine learning capability built into Qualcomm’s Snapdragon 835 Mobile Platform, which powers cutting-edge devices such as the Samsung Galaxy Note8, runs on an architecture designed to be fast enough to handle 5G Gigabit download speeds. In addition to enabling blazing-fast video and audio streaming, this superior processing speed also allows for AI-dependent applications such as real-time virtual reality immersion, natural speech recognition, filtering out background noise for better audio quality and eye and face recognition for security authentication.
Making Gaming Smarter
Gaming has always been a pioneer medium for tech innovation, and machine learning is no exception. One major application of machine learning is making gaming smarter by empowering games to adapt to the styles of players.
Game engine software provider Unity Technologies has positioned itself at the forefront of this field by announcing the release of Unity Machine Learning Agents, an open-source software code that can be used to link the company’s gaming engine to machine learning programs such as Google’s TensorFlow. This can enable non-player characters to learn from play through trial-and-error and develop strategies superior to those that could be programmed.
The platform’s applications aren’t limited to games either. Unity Machine Learning Agents can also be applied to robots to speed up their learning curve by allowing them to practice navigating a virtual environment.
Automating Transportation
Unity Machine Learning Agents is also designed to assist the development of autonomous vehicles, another major application of machine learning. Virtual gaming environments are easier for developing autonomous vehicle technology because VR objects are pre-defined so that the software already recognizes them, whereas real-life objects such as cars, pedestrians, and road signs have to be photographed and labeled so that the software can develop techniques of recognizing them.
By using virtual environments to train autonomous vehicle software, developers aim to equip machine learning for the task of handling autonomous vehicle navigation. In a real-world environment, autonomous vehicles use machine learning to integrate and analyze input from a wide variety of sources, including internal vehicle sensors, external cameras, radar, lidar and the Internet of Things. Mountain View, California startup Drive.ai is using machine learning to teach autonomous vehicles to adapt to driving under difficult conditions such as darkness, rain and hail.
By applying machine learning to mobile devices, gaming and autonomous vehicles, developers are paving the way for other AI innovations that will build on these breakthroughs. As machine learning becomes standard in smartphones and cars, it will become increasingly integrated into the Internet of Things and everyday applications, fueling competition and inspiring greater innovation.
