The 21st century is all about data, and IoT devices excel at producing it. The world has generated more data in 2017 than in the previous 5000 years added together. When you feel your data analysts cannot handle the amount of data, or that their analysis is one step behind, machine learning may be your way out of a tangled data mess.
A Century of Data
When it comes to the usefulness of data, there is never too much data; however, when it comes to storing that data or processing it, the more data you collect, the more difficulties — IoT devices excel at producing vast chunks of data. Some devices or sensors produce data every second and even every millisecond: multiply that by thousands or millions of devices, and you quickly drown in a sea of data.
Big Data and IoT Devices
Most predictions for 2018 say machine learning will go beyond the lab this year and, by processing the data found in the IoT, machine learning will turn data into actionable insight.
The need for Machine Learning in an IoT based environment was born as decision makers are overwhelmed not only by the amount of generated data but by the necessity of reacting instantly to the surroundings. CEOs, Data Scientists and Data Analysts usually look at the collected data to draw conclusions and make forecasts — but IoT data is just too fast. By the time the data scientist arrives at work and manages to finish the analysis, the situation may have already changed, and the actions suggested by data a couple of hours old may be obsolete.
Enter machine learning, the IT “sub-branch” to take over this job.
The Big Data Virtual Assistant
Imagine a flow of data coming in at high-speed, from various devices. All that data, commonly known as sensor data if IoT devices generate it, is being cascaded into a data lake, and the data scientist needs to overwatch it, process it, and analyze it. To act immediately to what is happening, you need fast response time to change. To react, you need to know, and knowledge in real-time is something that an AI can currently have slightly quicker than humans, as such, you could use the help of a Big Data Virtual Assistant. However, the lack of intuition, context understanding, and correlations make AI limited. Singularity may not be so far away, but we are not there just yet.
In this context, the practical approach for AI & Machine Learning in business is still limited, and can be viewed as a bonus, but not as a replacement. With the multitude of data coming in from various IoT devices, Data Scientists can be overwhelmed with the information they need to process. The advantages of an AI, in this case, are straightforward: fast response time. A data scientist can survive without AI or Machine Learning, but it is like surviving without a smartphone in 2018. We can, but do we want to? As such, you can view AI as being your virtual assistant that complements your Data Science department.
With sensor data, the most common type of data generated by IoT devices, a frequent task is to detect anomalies. Anomaly detection refers to the problem of finding patterns in data that do not conform to expected behavior. Anomaly detection attempts to quantify the usual or acceptable behavior and flags other irregular behaviors as potentially intrusive, as stated in the book Anomaly Detection with Machine Learning by Hanna Blomquist and Johanna Möller.
With Machine Learning, one can determine if a data instance is correct or not with the help of data-driven prediction. It can also be formulated as optimizing a performance criterion by making predictions from models built on knowledge of past data. Machine Learning is about learning, adapting, recognizing and optimizing – in short, to be intelligent.
As such, instead of reacting late to changes that could be vital to your business, your AI-robot friend can inform the management teams instantly of any changes, and action can be taken immediately.
We do envision a future in which smart robots may replace humanity, but while we are working towards the singularity, AI can be regarded as our robot colleagues that help us make faster decisions, at least when it comes to analyzing sensor data, or another type of real-time generated data.
AIs are built with a purpose, and most of them are programmed to achieve that purpose alone. Just because your AI is good at spotting anomalies, it does not mean it will correctly set your medical diagnostics. As such, in the case of IoT, AI & Machine Learning, the value proposition sits within the fast access to information, while the Data Scientist, the CIO, or CEO are still responsible for the decision making process. If further help is needed for making the right call, someone can surely build a dedicated AI for that, but never ignore your intuition and power of deduction from context. That’s something an AI can’t achieve yet.
Conclusion: AI & Humans working together for a brighter future
Robots and humans can work together in harmony, and we can set aside, at least for a while, the apocalyptic vision of AIs taking over the world and our jobs. If we look at the present, AI & Machine Learning helps us get faster responses, quickly make correlations, get answers to questions we have not even thought of asking years ago. Overal, it simplifies our lives in both terms of business decisions, as well as a way of life.
Detect anomalies, spot patterns, see the future. It is no longer magic. It is Big Data.