Abstract

Making a computer learn from data, instead of explicitly programming it to perform certain tasks, is a useful ability. It allows us to find – or at least approximate – solutions to complicated problems. This includes tasks which machines usually struggle with, or even ones that we as humans still do not fully understand.

For such a powerful tool, the basic principles of machine learning are surprisingly easy to understand. In this talk, we are going to take a look at the basics: what is a hypothesis, how to represent and use it, and how to pick the best one. We shall also see example algorithms, including linear and polynomial regression, naïve Bayesian classifiers and logistic regression. Lastly, we shall learn how to start writing programs that use machine learning approaches.