Bayesian machine learning

In essence, Bayesian means probabilistic. The specific term exists because there are two approaches to probability. Bayesians think of it as a measure of belief, so that probability is subjective and refers to the future.

Frequentists have a different view: they use probability to refer to past events – in this way it’s objective and doesn’t depend on one’s beliefs. The name comes from the method – for example: we tossed a coin 100 times, it came up heads 53 times, so the frequency/probability of heads is 0.53.

For a thorough investigation of this topic and more, refer to Jake VanderPlas’ Frequentism and Bayesianism series of articles.

http://fastml.com/bayesian-machine-learning/

Advertisements