Machine learning is a computer’s way of learning from examples. It’s a type of artificial intelligence, and will be among one of the big technological disruptions of the coming years.
While it isn’t a new concept, it is now evolving faster than ever before.
One of the first examples of a machine learning on its own was in 1956 when Arthur Samuels taught a machine to play checkers using a computer program. He had the computer play checkers against itself thousands of times, looking for strategies in the process.
“Machine learning has a really long history,” Data strategist at USF Jeremy Howard said at the CeBIT technology conference today in Sydney.
Since then, machine learning has become hugely complicated and technologically advanced. Google has developed a car which has driven itself over 1 million miles without an accident. Insurance companies can now predict insurance losses with almost pinpoint accuracy using various algorithms.
“There are extraordinary commercial results for machine learning,” Howard said.
“Just ten years ago experts were writing papers about how we would never be able to program a car to drive on the road because the rules are just too complex.
“They [Google] used machine learning very heavily to basically find out from the data: What’s a tree? What’s a pedestrian? What’s this car going to do next? How should I respond?
“Machine learning is actually letting us create thing that just ten years ago we thought we might never have been able to create.”
The acceleration of machine learning is going to bring about another revolution similar to the industrial revolution, Howard predicts.
“We’re hitting the point where we’re nearly as good as humans with machines,” Howard said.
“What happens in five years time when suddenly computers are ten times better? What does that mean for employment? It probably means we’re going to see something again like the industrial revolution where for decades afterwards we’re going to see massive social disruption.”
Machine learning has advanced so much in the last decade it’s difficult to determine what’s going to be possible in the future. But Howard outlined three concrete ways machine to machine learning will change the world we live in by understanding humans.
Scientists are already training computers to read breast cancer pathology reports which take about four expert pathologists to read, understand and reach a decision about which areas to treat.
“A machine learning algorithm was trained to automatically identify these areas and when they benchmarked them they found out it was more accurate then the best of the human pathologists,” Howard said.
“The idea of looking at images and understanding something particularly at a level beyond exerts is something traditionally computers could never do and yet now, at least in that area of research, we’re at a point where computers can work better than the best people.”
Computers learning to read
Computers haven’t been able to read and understand the subtleties of human language.
But this is changing, with a study coming out of Stamford that has created an algorithm that is teaching computers to read language.
“A Stamford researcher has been able to come up with a machine learning algorithm that was able to uncover the patterns [of human language],” Howard said.
“The accuracy of this algorithm is only about 5 per cent less than the agreement of humans with each other.
“This is now approaching human levels of ability to understand the sentiment of natural simulation.”
Computer object recognition
In a recent study thousands of images and thousands of sentences were put into an algorithm and a computer was able to match the picture with each sentence describing it.
“This algorithm is close to human level of performance with coming up with appropriate captions,” Howard said.
“We can now not only understand images and read text we can actually bring the two together.”