This month we'll focus a bit more on the 'programming' part of Probabilisitic Programming. First we'll have:
Probabilistic programming from scratch
Real world observational data is always imperfect or incomplete in some way. Those limitations mean that what we learn from our data is somewhat uncertain. We want to fill in the blanks to the extent possible and be able to say how confident we are as we do this. This is inference. Probabilistic programming makes it easier to learn from data. Let’s see how to build a basic probabilistic programming system from scratch in Python to introduce Approximate Bayesian Computation (ABC), which is a specific, extremely simple algorithm to perform Bayesian inference.
Bio: David is a Business Analyst and Software Developer with six years of corporate experience in building databases, managing and analyzing large data sets, and optimizing systems and processes. Machine Learning enthusiast focused on Natural Language Processing (NLP) and visual recognition.
Then we'll introduce and discuss "An Introduction to Probabilistic Programming" https://arxiv.org/abs/1809.10756 and explore some of the concepts we've covered using Pyro http://pyro.ai