AI and IoT: Taking Data Insight to Action
AI and IoT: Taking Data Insight to Action
This in-depth look at the state of AI and IoT explores how these technologies synergize with each other and what to expect in the future.
Join the DZone community and get the full member experience.Join For Free
Recent Gartner estimations lead us to believe that up to 20 billion connected things will be in use by 2020. Data is the oil of our century — but should we be concerned with a "data spill hazard"? Will artificial intelligence curb this threatening phenomenon, or rather, will it reveal the full potential of IoT data value?
If my calculations are correct, when artificial intelligence hits the Internet of Things... you're gonna see some serious sh*t.” — Doc E. Brown (or something like that)
After Big Data, What Will Be the Next (R)evolution?
The question is no longer whether companies should embrace big data analytics technologies. The answer is that they definitely should. Whatever the source we trust, we all know that the amount of data we are dealing with is growing very fast. See, for example, what happens online in 60 seconds. That's why IDC has stated that “over the next three to four years, digital transformation efforts will no longer be 'projects,' 'initiatives,' or 'special business units' for most enterprises.” Frank Gens, IDC SVP explains that these digital transformation efforts will actually become the core of their business. Remember former Jeff Immelt GE CEO quotes, “If you woke up as an industrial company today, you will wake up as a software and analytics company tomorrow.” The more data we generate, the more opportunities companies have to analyze and take advantage of it.
We are evolving in a technological era that experiences constant change, driven by rapid advances in research and developments, where industry standards and rules are getting more and more complex, customer demands are more sophisticated, and product lifecycles are dramatically shrinking. The top 10 big data and analytics predictions for 2017 include analytics becoming decentralized and moving to the cloud, as well as more sophisticated platforms emerging. The research states that “artificial intelligence investments will triple as firms begin tapping into complex systems, advanced analytics, and machine learning technology.”
Forrester Technology predictions anticipate that the next technology revolution may serve to create extremely different digital experiences and offer an opportunity to bring reality into predictive analytics and, more importantly, to support technologies that will drive new levels of speed and efficiency.
Among the few future-proof technologies that have the capacity to help enterprises to be able to gain the maximum potential from this revolutions, Internet of Things and Artificial Intelligence appear out way ahead of this game.
IoT Is Everywhere
So far, the most popular IoT application is the smartphone. There are approximately 3 billion smartphones being used in the whole world today and it should double that by 2020. As time passes, technologies and markets are growing and maturing in such a way that automotive systems, smart meters, security cameras, smart sensors, smart buildings, and whatever-smart-devices are becoming an integral part of our daily life.
Some Everyday Life IoT
The first connected thing was set up even before the invention of the World Wide Web. Read here about the story of the famous Internet Coke Machine created in 1982 by Carnegie Mellon University students.
Additionally, the Nest Learning Thermostat is one of the most trendy smart home devices. It learns what temperature you and your family like and adjust the temperature. You can edit schedules, thanks to an app, and you can also receive alerts. Smart lighting systems are also easily finding their way into our homes. The Philips Hue lighting system, for example, claims that not only it provides you with a fun app to manage your lights but also helps you reduce costs and save energy.
The list is long, and I could go on with smart cameras, motion detectors, pet feeders, toothbrushes, key locators, connected fridges, smarts door locks, shared bikes… and in a few paragraphs, we will talk about Industrial IoTs.
But first... what do shark recognition, gunfire detection, and bike sharing have in common?
Whether or not data coming from everyday life objects offers great importance and value is a good question. As Bill Schmarzo (CTO, Dell EMC Services) rightly stated, “‘Connected’ does NOT necessarily mean smart.” How do you make the best of it? Who can make the best of it?
Three examples can be given for illustration. One in China, related to the bike sharing sector, one in the USA within urban lighting market, and the last one in Australia in the area of shark detection.
In China, the oFo bike app provides you with the code to unlock a shared bike and handle the payment process. That helps to answer the questions: How many people use the bikes? Where do most people take or drop bikes off? In the same way as oFo’s people use the data to optimize their distribution plans, the data is gold for city officials in order for them to plan the design of their urban architecture: roads, bus lanes, cycle lanes, walkways etc.
In California, GE is working with ShotSpotter, which is designed for detecting and locating gunfire in real time. Teaming up with Intel and AT&T, GE is combining cameras, microphones, and sensors into its intelligent LED streetlights. GE is then upgrading its urban lights far beyond simple illuminating capabilities. While today, ShotSpotter is using complex real-time algorithms and the work of a team of human specialists, the data collected by IoT in real time can be associated with available historical data to be processed by Deep Learning for further automation.
In Australia, the University of Technology Sydney (UTS) is harnessing Deep Learning and Artificial Intelligence algorithms to detect sharks from drone footage. First, they preprocess public videos of sharks. This is the learning part. Then, a Neural Network runs detection and recognition algorithms. Empowering the action of drones, this provides a real-time search and rescue service.
IIoT: Industrial IoT
When it comes to considering the industrial aspect of the Internet of Things, we immediately think about the most advanced sectors including healthcare, automotive, etc.
In their article Preparing for the Challenge of Artificial Intelligence J. Geetter and D. Van Demark give a portrait of the situation with regard to the challenge of AI in healthcare. According to them, Artificial Intelligence helps to address issues including precision medicine, predictive analytics of patient populations, "clinical decision support," cybersecurity, and patient engagement. They are specifically identifying that the combination of AI and IoT might be designed to deal with the creation of integrated health systems where wearable and remote monitoring devices could provide the patient with the ability to self-diagnose.
Healthcare-oriented smart devices are numerous. They may help in measuring your sleep, monitoring your heart rate or blood pressure and rhythm, oxygen saturation, systolic pressure, tracking your weight and body fat percentage, and also measuring your body temperature or stress level. Thus the promises of AI in the healthcare are impressive. At Stanford University in California, E. Ross, a vascular surgeon, finds that Artificial intelligence may help prevent heart failure.
When you associate automotive and IoT, you then step into the Industrial Internet of Things. This line of business is an early adopter. We've heard this for a number of years, since car makers began collecting data from the sensors with which they equip modern vehicles in their goal to move from corrective maintenance (when a piece is already broken) to preventive maintenance (when you replace it before it fails), then to predictive maintenance, when you are able to determine when it will fail.
There is the real progress. When the replacement of any piece of your vehicle is based on global statistics, in most of the cases, it’s a waste of money, as it’s changed too early without taking into account the specificity of your car or the way you use it.
Then, the purpose of predictive maintenance is to more accurately evaluate the right time to replace the part of your vehicle, maximizing its life cycle while reducing the risk of seeing your car break down. It's all about collecting data from you vehicle, analyzing it, confronting it with historical data, building a predictive model, and ultimately — this is exactly where Machine Learning comes into play — running algorithms that will calculate the right time to proceed with the replacement of the identified piece.
That said, two things to keep in mind: First, predictive maintenance is also a good way for companies to gain competitive advantages by optimizing their maintenance services. Second, this applies to other industries, including machine tools, helicopters, aircraft, aerospace, wind turbines, and, hopefully, soon our washing machine at home.
Get a “Lethal Weapon” by Combining IoT and AI
We know that Artificial Intelligence is the new cornerstone of Big Data Analytics. But it’s difficult to anticipate the extent to which the Internet of Things can be engineered with Deep Learning, Machine Learning, and Artificial Intelligence, as part of the Artificial Intelligence Ecosystem, to come up with enhanced capabilities.
Size does matter. Currently, we produce 2.5 exabytes of data every day. According to IBM, “Globally, the data created by IoT devices in 2019 will be 269 times greater than the data being transmitted to data centers from end-user devices.” Extracting information from data and then engaging valuable actions from these insights was always a challenge. So, how do we handle the overflowing flux of data generated by IoT?
Time is key. The advent of the IoT age speeds up the need for fast processing. Not only do applications and customer requirements imply a high level of speed, but also, in many cases, they definitively require real time. Think about the shark recognition program or self-driven cars as examples.
“Data is only useful if it is actionable. And to make data actionable [...] is about ‘connected intelligence' — which is where AI and smart machines come into the equation," said David Stephenson.
Based on their context, their accessibility, and their accuracy, whether they are structured or unstructured, data must be actionable. The deluge of data generated and transmitted from IoT must lead to actions or decisions. In his article IoT and AI: Made for Each Other, D. Stephenson emphasizes that IoT and AI are mutually impactful. He explains that Artificial Intelligence strengthens the Internet of Things by enabling predictive, prescriptive, and adaptive analytics while IoT brings its real-time capacity to AI. He predicted that any connected thing will be able to become intelligent thanks to Artificial Intelligence.
IIoT Data Integration Challenge
When considering a high volume (up to millions of connected devices) of complex data in motion produced from sensors placed on high tech machines such as jet engines, traditional data integration tools might not do the job. In these cases, GE Digital is leveraging the power of Machine Learning to operate data cleansing and data integration tasks based on a semantic data models. They explain that their architecture “was able to integrate 52 different data sources in about 10 days, whereas traditional ETL processes would have taken more than a year to complete”.
The model underlying the architecture deals with operational controls, sensor channels, business information, environmental data, and geospatial data. The system, which is using supervised and unsupervised learning methods, is able to handle a broad spectrum of problems, including out-of-sequence data streams, delayed streams, data gaps, and signal noise. The ultimate goal of this ML-boosted platform is to handle data that was never seen before without having custom decoders.
I guess that after everyday objects and industrial objects, the ultimate dream would be to connect a human brain to the internet. Whether that is a bit optimistic or it can be effectively done, a press release states that a team of researchers at Wits University in Johannesburg claims that they made it by converting electroencephalogram (EEG) signals (brain waves) in an open source brain live stream.
If this was actually working, it would enable amputees and physically disabled and paraplegic citizens to recover some use of their bodies, proving that technology could definitely help to make the world a better place.
One more thing... I cannot end this article without quoting this comment I read, related to the 'Brainternet' project: “If they could only figure out a way to get people to USE their brain on the internet.”
Opinions expressed by DZone contributors are their own.