What is Machine Learning (ML)?

 

The first time I saw a dog, I didn’t know what it was, but someone told me, and I knew to assign this label to all future dogs. Simple. Well not really, as there are many types of dog. Unconsciously I took note of the number of legs, coat type, ears, head shape, etc., and using these additional labels, I was able to identify virtually all dogs, regardless of breed. In addition, I was able to distinguish similar animals, e.g. cats and dogs, horses and donkeys. I could even subdivide one species into subsets, e.g. dalmatians, terriers, spaniels.
 
This type of thinking isn’t limited to humans; other animals can distinguish predators and prey, good plants from poisonous, but it is a highly complex activity and it has taken millions of years of evolution for our brains to be so adept, and to be able to make judgements on new animals based on our existing knowledge.
 
Computers are not innately able to ‘think’ in this way. However, machine learning, which is part of artificial intelligence (AI), is the science of making machines learn and think like humans. This can be accomplished in a number of ways, for instance using neural networks or algorithms that allow computers to learn from data without explicit programming. So why get a machine to learn? The answer is that computers can process innumerably more data than a human in a consistent and logical way, and this allows us to investigate problems which were previously thought too complex.
 

 

 

Click here for our next Disruptive Technology Explained blog on Artificial Neural Network.