by Hassan Faouaz –
It’s no secret that great strides have been made in information technology in recent years.
Instantly transport an IT professional from 20 years ago to the present day, and that person would likely be astonished at the progress that has been made — particularly with regards to our ability to gather, store, and process massive volumes of data.
But transport any one of us from present day to only two years into the future, and it’s just as likely that we’d be equally astonished at the progress made. What would we find so astounding? It’s probable that we would be amazed at the strides made in deriving value from Big Data.
I believe that we’re standing at the dawn of a new age. It’s an age in which Big Data will combine with machine learning and closed-loop analytics to enable incredible advances. The analytical insights made possible will benefit business in a big way.
And those benefits will filter down to positively impact each of our lives daily.
Closing the Loop
Flowcharts aren’t used as much these days as they once were, but it’s a good tool for illustrating the difference between open-loop analytics and closed-loop analytics. In a flowchart, you have a logic flow that inputs into a decision (remember the triangular symbol representing a decision module?), and an output from the decision.
An example might be a person sorting through tweets to determine whether they’re positive or negative. He picks up the printout of a tweet, considers it for a second, places it in either the positive or negative stack, and moves on. An open-loop process.
What have we learned from this open-loop process? We’ve learned whether a single person has expressed a positive or negative opinion in a tweet. And that’s all we’ve learned.
But what if a machine is performing that process?
Machines are now capable of determining the sentiment of the tweet — positive or negative — just as well as the human. But the machine is capable of gleaning additional insights and information from the tweet and feeding that information back into the loop. And that information builds and is used to further refine the sorting process that helps to tease out additional analytical insights.
The machine is learning. And what it learns is fed back into the loop, and used to continually tweak and improve the process. It’s now a closed-loop process.
The Future is Now
The above is a simplistic example of closed-loop, machine learning analytics. But it’s not a futuristic example. The technological capability exists right now for machines to learn on the fly, and to feedback what they learn to continuously tweak and improve a given process.
And that process might be far more impactful than simply analyzing tweets.
Researchers at the University of British Columbia, for example, have been experimenting with a closed-loop system that relies upon reinforcement learning in the autonomous administration of anesthesia.
The machine monitors a variety of biometric markers, such as blood pressure, blood oxygen levels, and even brainwave activity, and determines — without human input — when and how much to alter the amount of anesthesia delivered to the patient. The system has been used to induce and safely maintain sedation in hundreds of patients undergoing surgery.
How effective is this closed-loop, self-learning system? “We are convinced the machine can do better than human anesthesiologists,” offered one of the project team members, as reported in the Washington Post1.
The automation of anesthesia is not a new concept; it’s been around for decades. But an autonomous, self-learning, closed-loop system is a new concept, made possible through the convergence of Big Data and machine learning. And it’s just one of countless closed-loop, self-learning applications that will astonish us all in the coming years.
Another example might be driverless cars. The cruise control in your car requires that the speed of your car is continuously fed into your car’s computer. Each input of measured speed triggers a decision to either increase or decrease speed, or do nothing. That’s open-loop.
But completely automating the process in creating a driverless car (Google’s Waymo; Hyundai’s Ioniq; Tesla; etc.) requires a self-learning, closed-loop system.
Early Benefits for Pioneering RCG Customers
At RCG Global Services, we intend to be right at the forefront of this exciting new frontier. And we’ve already seen some impressive results from customers that are early adopters of machine learning and closed-loop analytics.
Here are a couple of examples…
- A global manufacturer of automotive batteries embeds sensors in their batteries. The sensors report a range of metrics that is constantly analyzed, and the results fed back into the system to continuously tweak predictive analytics about the future performance of the battery. As a result, customers can be informed before the fact when their battery is likely to fail. (Sure beats finding out your battery is dead in an isolated parking lot in the middle of the night!)
- A retailer has begun to use closed-loop analytics to monitor tweets real-time, specifically for the purpose of determining when to open additional checkout lines. Tweets from customers complaining about the length of their wait prompt a recommendation to management to open new checkout lines. And when customers are zipping through the checkout lines without delay, the algorithm can recommend that a register or two be closed, and the idle employees assigned to more productive chores.
We’ll all benefit from the amazing technological advances that will be enabled by utilizing Big Data in closed-loop analytics. But some companies will benefit more, and sooner, than others.
We’re working hard to assure that many of the companies that populate that ‘more and sooner’ group will be RCG Global Services customers.
#IdeasRealized
Works Cited
1. Washington Post (2016, May 15) Retrieved from https://www.washingtonpost.com/news/the-switch/wp/2015/05/15/one-anesthesiology-robot-dips-its-toes-into-whats-possible-this-one-jumps-all-in/