6 ms·
Answer is not so black and white as everything in ml has to use probability. You can ignore this unless you are among 20 top researchers who are working on fron
by master_yoda_1 6y ago
Answer is not so black and white as everything in ml has to use probability. You can ignore this unless you are among 20 top researchers who are working on frontier of ml. Bayesian probabilistic techniques does not work or are very slow for any practical purpose.
- amkkma 6y agoBut does it aid in understanding regular models, as they might have a bayesian interpretation?
- master_yoda_1 6y agoYeh for sure, but it's an overkill. It's like reading quantum mechanics to understand Newtonian mechanics. If you want to get a feel of bayesian ml here is an easier book, "Regression and Other Stories" https://avehtari.github.io/ROS-Examples/ https://avehtari.github.io/ROS-Examples/
- amkkma 6y agoThanks. How's that relate to Gelman's BDA or statistical rethinking?
- master_yoda_1 6y agoBda is more advanced version of this book. Stats rethinking is also good and one can start from scratch as it has more code and less maths.
- dklend122 6y agoTy