How does machine learning work?
This may not be one of life’s big questions. But the topic of machine learning is all the buzz in the artificial intelligence (AI) community these days. The idea that machines are actually learning might be news to you. And you might wonder how they do this without a brain. Yes, machines don’t have brains. Not like human brains, at least. But they are learning.
So Then How Does Machine Learning Work?
The notion of machines actually learning might be a little unnerving. Especially if you were raised on sci-fi books and movies. But machine learning is about as scary as a bobble head. It’s far more similar to data mining than it is to evil robots taking over the world. In both systems, data is searched in an attempt to find patterns. The difference is that with data mining, the data is extracted for the sake of human comprehension. Machine learning uses the data to find patterns and then adjust program actions according to those patterns.
And these machine learning algorithms can be used to either apply what has been learned in the past to new data, or to draw inferences from datasets.
Okay, so let’s put it in simpler terms. Let’s say that you want to buy the best cotton candy machine on the market. You find what looks like the perfect model. Then you search a little deeper and eventually land on the reviews for that perfect model. If words like “excellent,” “great” or “awesome fluffability” show up, you can feel confident about your move toward making a purchase. On the other hand, if words like “bad,” “poor quality” or “caught on fire” keep appearing, you know it’s probably best to move on to a different machine. Or even scrap the idea and buy something less liable to rot out your teeth.
In either case, the reviews help you take action based on the pattern of words that appeared in them. The buyers who wrote product reviews will influence other buyers. And their reviews will have an influence on future purchases.
From this, a pattern now exists across the people who already made a purchase and the future buyers of the product. Machine learning attempts to encode this human decision-making process into usable algorithms.
That’s the most simplistic answer to how does machine learning work.
It’s important to understand that three conditions must be met before one can apply machine learning to a problem though.
1. There must be a pattern in the input data to arrive at a conclusion.For example, if we believed that the reviews didn’t offer any meaning, then they wouldn’t help us to make a decision. For machine learning to solve a problem, the algorithm must have a pattern to infer from.
2. There has to be enough data to apply machine learning to a problem.If there were no product reviews at all, it will be difficult to arrive at a decision as to whether to buy the product. Right?
3. We, humans, are unable to formulate a mathematical expression that describes the behavior of the problem.This is the stuff that causes most human brains to implode.
So now machine learning is used to find meaning in the data and perform “learning” to come up with a mathematical approximation that describes the behavior of the problem.
That’s it in a nutshell.
In conclusion, machines are not taking over the world. Not yet.