pml2-book
pml2-book copied to clipboard
probml
Probabilistic Machine Learning: Advanced Topics
near the end of the 2nd line after equation (12.64) in "that have a constant energy, The trajectory", the comma should be a dot instead.
In the end of the second paragraph, "correpondence" should be "correspondence".
In the second paragraph, the second to the last line, "we shpw a grid-based", "shpw" should be "show"
On the (latex) page 816 of version 2023-08-15 mentions: "Indeed, these concerns are why Google has not (at the time of writing) released its version of ChatGPT, known as LaMDA"...
Equation 12.77 for the MH acceptance probability uses $-\mathcal{H}(\mathbfit{\theta}^\ast, \mathbfit{v}^\ast) + \mathcal{H}(\mathbfit{\theta}\_{t-1}, \mathbfit{v}\_{t-1})$ immediately after asserting that $\mathcal{H}$ (the total energy) is constant. Should this be the potential energy $\mathcal{E}$...
There is a typo in the exponent of the std in the FIM for the univariate Gaussian in eq. (3.2.34), the bottom-right element of the matrix should be: 2/σ^4 rather...
in the second to the last line of the section "enabling the us of more powerful optimization methods." "us" should be "use“
Hi Kevin, As mentioned in issue (https://github.com/probml/pml2-book/issues/253), the formula 24.33 need "+constant". But it seems that the last fix add "+constant" in formula 24.34 rather than formula 24.33 carelessly.
- In Figure 25.2, the right ) should be outside of {t+1} - In section 25.2.4, several \hat is missing. (1) "We usually set $\hat{\alpha}_t =$ ..." after eq. (25.26);...
The optimal critic for Pearson divergence should be $2 (r(x) -1)$, while the current expression is for Squared Hellinger divergence.