2024-11-25

Swap order of ch 6 (optimization) and ch 7 (intro to inference), as described at #351

2024-11-23

Fixed latest round of github issues, added a few new paragraphs and references here and there...

2024-06-26

• Fixed all known issues since last release in August 2023.
• Fixed broken cross references from supp2.pdf to main text.
• Added some new refs and a tiny bit of extra content here and there (eg sec 5.7 on algorithmic info theory)
• total number of pages has grown by 8 (to 1362), so page numbering is different (but section, equation and figure numbering is unchanged)

2023-08-15

This version fixes all known issues to date, and coincides with the official release of the hardcopy for volume 2.

2023-08-14

Fixed several small issues (see https://github.com/probml/pml2-book/issues?q=is%3Aissue+is%3Aclosed).
Just in time for the official release of vol 2 hardcopy :)

2023-04-01

Fix a few typos, add a few more refs, clarify a few sections, add a few more small subsections here and there (without breaking eqn or section numbering), etc.
(See github issues for changelog to book2.pdf; note that supp2.pdf has not changed since last release)

2023-01-22

• Cleaned up a few more figures
• Fixed a few more typos
• Moved the section on deFinetti from ch 2 (prob) to ch 3 (bayesian stats)
• Added a few more refs

Sent this to MIT Press

2023-01-19

Hopefully this is camera ready... :)
(Changes listed #209)

2023-01-16

• Still not actually quite "camera ready", but almost there...
• Small tweaks to pacify MIT Press
• Made various changes (listed at #209) to further improve the content, notably the coverage of frequentist statistics and variational inference
• Moved some stuff into online appendix to stay within page limit

2023-01-02

This is the "camera ready" version I am sending to MIT Press.

• made many small edits at the request of @mitpress, including adding 100s of "Oxford commas".

• completely rewrote ch 25 on diffusion models (now includes SDEs).

• improved ch 8 on Gaussian filtering/ smoothing (pseudocode now matches our dynamax library :)

• added short new section on hypothesis testing (sec 3.12)

• moved HMM forwards backwards into sec 9.2 (message passing on chains)

• moved some stuff from the main text to the online supplement to meet the page limits.
  Specifically, moved non-parametric Bayes (ch 31) and LVMs for graphs (sec 30.2) to online.
  Moved sec 3.2 (Bayesian concept learning) back into main.
  Moved sec 15.3.9 (logreg for Berkeley admissions) back into main.

• added more details on basics of probability theory (sec 2.1)

• tweaked presentation of particle filering resampling algorithms (sec 13.2.4)

• tweaked section on text generation with transformers (sec 22.4.1) to mention chatGPT

• fixed typos, and other cosmetic things (eg changed some chapter titles)