The deep learning revolution for artificial intelligence
How will neural networks aid facility management? RAE STEINBACH investigates.
Artificial intelligence (AI) has been a part of our lives to some small degree for many years now. However, older AI technologies were very simple and narrow in their range. An email spam filter is a good example of narrow AI; it uses a predefined set of rules to evaluate emails and determine whether the message should go to your inbox or into the spam folder.
Recently, we have started to see AI technologies that are much more advanced and capable of performing tasks that are far beyond the abilities of AI systems that existed just a decade ago. On-demand apps by brands like Spotify and Netflix are using the technology to make improved user recommendations.
Many of these changes to AI technology come from a resurgence in interest and research into deep learning. This technology is rapidly advancing the field of AI and, as a result, it is becoming an ever-present part of our everyday lives. This article will look at some of the ways deep learning is expanding and improving upon the AI systems that are changing how we live and the way in which business is done.
What is deep learning?
Artificial intelligence is a term that covers any machine that can simulate intelligent action. With simpler forms of AI, a programmer provides the system with a set of rules, and the machine follows those rules to perform a function.
Machine learning is an application that provides an AI system with the ability to learn from experience. Using statistical models and algorithms, the AI system can progressively get better at what it does without a human having to go back and program the AI to perform better.
Within the field of machine learning, you also have deep learning. Instead of the task-specific algorithms used in simpler forms of machine learning, deep learning uses neural networks. A neural network is a computer system based loosely on the way the human brain functions. In the neural network, thousands to possibly millions of processing nodes take on the role of neurons. As the network is fed training samples, it builds connections between the processing nodes. The connections can change in various ways as the system is exposed to new data. If a connection helps it perform a task better, it gets stronger, but if the connection leads to mistakes, it will get weaker. This is similar to the way in which the human brain builds connections between neurons as it learns from experience.
Developments in deep learning
At this point, most consumers have used products and services that have been improved by deep learning. The following are just a few examples of the ways deep learning is changing our lives.
Consumers can now buy a number of connected devices for use in the home. We have smart thermostats, smart lights, smart appliances and even smart security systems. By analysing data collected from a variety of sensors and devices, a smart home can offer a variety of convenience features while also saving you money and making your home safer.
If you use voice assistants like Cortana or Siri, you have probably noticed that machines have become much better at understanding human speech. This is due to deep learning being applied to the field of natural language processing. By training the software with millions of examples of human speech, it provides a voice assistant with the ability to recognise the many different ways in which a word can be pronounced.
One common example of image recognition is the way Facebook uses facial recognition to suggest tags for pictures. By analysing the faces in the photo, the system can find faces that match and it makes suggestions based on the results. Beyond that, image recognition has the potential to change the way diseases are diagnosed. Several studies have shown that AI systems can be trained to analyse medical imaging to reliably detect diseases. In some cases, these systems even outperform human doctors.
Deep learning is also becoming a critical tool for facility managers. With smart systems that can analyse data collected about processes in the facility, this technology is helping to improve efficiency and boost production.
As an example, smart sensors can now be placed in the equipment used in a facility. The data collected from these sensors is then analysed to predict failures and identify performance issues. With this information, a facility manager can get ahead of potential equipment failures and find ways to streamline processes that improve productivity.
Image recognition could also be used to improve security at sensitive locations. For example, facial recognition is a form of biometric authentication that could improve security while also being a more convenient form of access control. One good example of this is the facial recognition software that is already available from companies like FaceFirst.
At the current time, deep learning is very resource-intensive and requires a specialised skill set. A neural network is a massive computing system that is far more powerful than your average computer. As a subset within a subset, working with deep learning is generally beyond the skills of your average programmer.
Due to the high demand on resources and the scarcity of qualified programmers, the application of deep learning is beyond the reach of many businesses. That said, it is expected that this will change in the near future. With advances to the hardware and more programmers deciding to specialise in deep learning, it is only a matter of time before this technology becomes more accessible.
Rae Steinbach is a graduate of Tufts University with a combined International Relations and Chinese degree. After spending time living and working abroad in China, she returned to NYC to pursue her career and continue curating quality content. Rae is passionate about travel, food, and writing, of course.
Image: 123RF’s Konstantin Faraktinov © 123RF.com
This was originally published in the Apr/May 2019 issue of FM Magazine.