Machine Learning Applications
Machine learning (ML) is the capability of systems to automatically learn and improve from experience independently, without human intervention. Used with artificial intelligence, the capabilities of machine learning are endless.
Capitology Blog sat down with Dr. Robert Steele to discuss machine learning and its applications. Dr. Steele serves as Chair and Professor of Computer Science. He holds a doctorate in Computer Science, has authored more than 140 peer-reviewed publications, and his work has been patented and successfully commercialized.
“Machine learning and artificial intelligence can be applied to all sectors in thousands of different ways,” says Dr. Steele.
Some of the fields where machine learning has been heavily adopted include health care, transportation, finance, defense, and speech recognition/automatic translations. In fact, Google Translate, which uses the Google Neural Machine Translation system, is on par with human translators.
Dr. Steele adds that machine learning is about providing the capability to predict the future in a limited context. In general, by looking at large data of historical events you can predict the future for upcoming events.
“We did some work where for elective surgery patient admissions, we predicted in-hospital mortality,” says Dr. Steele. “Every year a fraction of patients that go in for elective surgeries do die. We were able to predict the death of the patient at the time of scheduling. You could apply that model at scheduling and if you ran that data and saw the patient could die could potentially prevent that outcome.”
Machine learning can also analyze CAT scans and x-rays to find diagnoses or predict the path of diagnoses. In the finance industry, for example, creditors can use machine learning look at a person’s history and determine the likelihood they will default on a loan. It can be used to detect malware, among many other cybersecurity and defense applications.
“Machine learning provides the ability to predict the future with a certain accuracy. The same techniques you use to predict the future are the same techniques you use to make judgements that only a human could do normally,” says Steele.
One example is autonomous vehicles, where the self-driving component relies on machine learning. You are training the computer in the vehicle to see an image in front of it and identify what it is – another car, tree, stop sign, etc., so that the vehicle knows how to react. However, you can’t guarantee that every possible scenario a car could encounter is going to be known by the autonomous vehicle’s system.
Machine learning works by seeing many past examples it can recognize future examples. This brings up the question of what happens if it sees something that it has never seen before. In a road setting, things are going to happen the system doesn’t recognize.
“You can apply machine learning to any sector and any problem, but it’s fully dependent on the data available,” says Dr. Steele. “As a student of machine learning you need to understand if you have the right data. You can’t do machine learning without the historical data and it has to be the right kind of data. It’s almost a creative activity. You have to understand with that data, I can answer that specific question.
Capitol Tech offers bachelor’s, master’s, and doctoral degrees in both computer science and data science. Both areas of study include coursework in machine learning. For more information, email firstname.lastname@example.org