When I google search for the term, only 6 pages showed up. There is no authoritative paper on this, except https://arxiv.org/abs/1306.6430 which argues for using informatics concepts to generalize a prior. It seems an attempt to reconcile Bayesian inference with machine learning concepts. Is bayesian updating framework a valid concept in the statistics community?
1
-
1$\begingroup$ If you're specifically interested in the intersection of Bayesian inference and machine learning algorithms, I recommend this paper by Khan & Rue. $\endgroup$– DurdenCommented Sep 12, 2023 at 21:09
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
You are confusing "general(ized) Bayesian inference" with "Bayesian inference." Users of the first are not generalizing priors, they are generalizing Bayes rule by allowing the likelihood to be replaced. It's still Bayes' rule that is being used to perform inference on unknown quantities.
I don't understand why you assume there needs to be reconciliation between these two concepts. They aren't at odds. There is Bayesian machine learning, and there is machine learning that is not Bayesian. The difference (in my opinion) is whether practitioners use priors or not.
To answer your final question: yes, I believe so.
Welcome to Crossvalidated!
