As the “Internet of Things” is gaining popularity in the world of technology and the most high-profile technological words of the year, the debate about how to make it work is heated. The Internet of Things will produce a large amount of data - data that will help cities predict emergencies and crimes; will give doctors real-time information from pacemakers or biochips; optimize performance in various industries; provide an important connection between self-driving cars; to ensure the operation of smart homes with connected home appliances.
The possibilities offered by the Internet of Things are truly endless. But there are questions.
As the number of devices and sensors connected to the Internet of Things continues to expand, the amount of data grows to mind-blowing levels. This data contains valuable information — what works well and what doesn't — indicates conflicts and opens eyes to new opportunities and connections in different segments.
Sounds great. But the big problem remains finding ways to analyze this flood of data. If you have ever tried to find a connection in terabytes of machine data, you know how difficult it is. People are simply not able to study and understand all this data — and with traditional methods, even if you reduce the sample size, it takes too much time.
For the Internet of things to take shape, as promised, you need to increase the speed and accuracy of analyzing large amounts of data. If this is not succeeded, the consequences could be catastrophic, from irritation — when household appliances do not work together as planned — until life threatens — when hundreds of cars do not behave according to plan.
The only way to keep up with the volumes of data generated by the Internet is to incorporate a hidden understanding through machine learning.
Machine learning is already in use
Machine learning is “a subfield of computer science and artificial intelligence, which is engaged in the construction and study of systems that can analyze data not only according to pre-programmed instructions.”
Although this definition sounds like science fiction, it is already present in everyday life. Pandora uses machine learning to determine which other songs you might like; Amazon.com does the same with books and movies. Both systems are built on what they learn about the user and refine over time, learning more and more about their behavior.
In the case of the Internet of Things, machine learning will help companies take billions of data points and weld something meaningful out of them. The general principle is quite simple: you need to view and analyze the collected data in the search for patterns or similarities from which you can extract something and then make the best decisions.
For example, wearable devices that track your health are still part of a growing industry - but soon they will become devices that are interconnected with the Internet and will monitor your health in real time.
The goal is that your doctor will receive notifications if certain conditions are met - your heart rate rises to an unsafe level or even stops, for example. To be able to identify potential problems, data must be analyzed in terms of what is normal and what is not. Similarities, correlations and deviations should be quickly identified based on real-time data flows. Can a person do this physically? View thousands of patients in real time and pinpoint when to send an ambulance? Hardly.
To analyze the data immediately after collection, accurately defining new and already known patterns of behavior, the machines also need to know the normal behavior of each patient, as well as the critical level of health behavior.
The realization of the Internet of Things depends on whether it is possible to penetrate the essence hidden in the increasing sea of available data. Since the approaches currently do not scale to the volume of the Internet of things, its future depends solely on machine learning, which can find patterns, correlations and anomalies in the data. If it succeeds, it will improve almost every aspect of our daily life.
The article is based on materials .
Comments
Post a Comment