I'm currently struggling with understanding the Bayesian approach to machine learning. Which is one of the paradigms presented in Bishop Pattern Recognition and Machine Learning. Since some parameters that influence the posterior distributions such as the training points t in formula (3.49)

$p(\mathbf{w} \mid \mathbf{t})=\mathcal{N}\left(\mathbf{w} \mid \mathbf{m}_{N}, \mathbf{S}_{N}\right)$

aren't really random variables I was wondering whether we can say $p$(t) $= 1$ for instance.

  • $\begingroup$ Your $\textbf{t}$ is not a parameter but as you said your data points. In Bayesian inference, the data contribution is expressed through the likelihood function i.e (3.10) in the book. So I don't think that you can say something as $p(\textbf{t})=1$. $\endgroup$ – Fiodor1234 Dec 10 '20 at 17:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.