Traditionally a computer can only do what it’s been taught to do. Machine Learning challenges this assumption by teaching computers to look at the information they’ve been given and learn from it. Machine Learning algorithms have been around for decades, but are pretty taxing on hardware and haven’t been widely used until recent advances in computing made them more practical. The most common ML algorithm is the neural net, which simulates the way the human brain looks at information. The neural net exaggerates similar data so that it becomes different enough that the computer doesn’t confuse it. Another common ML algorithm is the decision tree, an algorithm that constructs a virtual model based on previous results to help the computer make connections between a cause and it’s effect. These algorithms don’t exactly simulate the way humans make decisions, but they come pretty close.
A decision tree that a ML algorithm is using to predict if someone survived or died in the Titanic crash. The computer keeps track of the likelihoods of someone surviving or dying given certain facts about that person.