Learn ML and AI

New to machine learning and AI? Our collection of tutorials, blog posts, and videos will help you get started with ML.

Check out learn.spell.run for more.

Machine Learning 101

Machine Learning is a technique within computer science for constructing algorithms from data. Instead of explicitly writing the logic in code, a model can be created to accurately recognize patterns from within a dataset that it's been trained against. Traditionally, machine learning methods have been very task specific, requiring domain knowledge expertise by the algorithm's creator.

Recently, the most popular approach within machine learning has allowed a much more generalizable solution, called neural networks. Neural networks simulate in software the way neurons connect in the brain.

You can think of these networks of connections as being separated into layers. The more layers that are added to the model, the more accurate and sophisticated the model becomes. This is why training neural networks is sometimes called deep learning.

However, training deep learning models can take significant amounts of computation. It can be done on a typical home computer, but the time it would take could be prohibitively slow. Luckily, modern hardware, specifically with GPUs (Graphics Processing Units) offers huge gains in speed. With GPU machines, a model that may have taken days to train now takes hours.