Luchezar Petrov

16.04.2021 (Friday) 19:00 EEST

In this report we showcase the application of Bayes theorem in the process of machine learning. Bayesian methods allow us to add prior knowledge to our machine learning model using the so-called prior probability. This means specifying probability distributions over the parameters of our model. That way we could combine the objectivity in our observations with the subjectivity of our prior knowledge. The result is a probabilistic model, which has associated probability distributions to its trained parameters. This way we could assess the uncertainty in both the parameter estimates and the model’s predictions. The concepts will be illustrated using the programming language Python and the package PyMC3.

Share This