bayes-hyperparameters-distinct-from-parameters
IN premise — summaries/2026/08/24/wiki-Bayesian_inference-chunk-1.md
Created 2026-08-25T02:58:42+00:00
Hyperparameters (α) parameterize the prior distribution itself, distinguishing them from model parameters (θ) that the posterior is estimated over.
Summary
The system must treat two kinds of numbers as doing different jobs: one set (hyperparameters) sets up the starting assumptions about what's plausible, while the other set (parameters) is what actually gets updated as evidence arrives. Keeping these roles separate is essential because you choose or tune the first set, but you infer the second from data — mixing them up would collapse the distinction between your prior choices and your learned conclusions.