Probabilistic modelling with Scala
summary
Demonstrates building Bayesian networks in Scala with Scala + base gala, constructs a student model and a Monty Hall model, and runs concrete inference. Shows Monty Hall yields Monty Hall 2/3 result (switching gives two thirds), uses CPTs with a 30% prior for “smart”, and reports numbers: strong recommendation rises from ~50% to 76% when intelligence is observed, then falls to 63% if exam difficulty is hard; observing a strong letter gives 45% smart, adding a high test score raises that to 93%. Mentions EM for parameter learning and tradeoffs of exact, BP, MCMC and variational methods.
AI-written summary · about
chapters (10)
- 0:00Intro
- 1:22Bayesian networks
- 4:40Why should you care
- 6:19Disadvantages
- 8:52Examples
- 12:01Monty Hall Problem
- 16:39Bayesian network life cycle
- 18:17Bayesian networks for prediction
- 26:17Algorithms
- 28:31Additional material
related talks
42mBayes and Bool Walk into a Bar
29mProbabilistic Programming: What It Is and How It Works
29mMotivating Probabilistic Programming
43mProbabilistic Programming in the Real World
32mStructured Deep Learning with Probabilistic Neural Prog
42mProbably
55mHigh-performance functional bayesian inference in Scala
43mHow to look like a statistician (...)