2022-06-28

• Major refactoring: split the chapter on message passing into 2 chapters, one focused on state space models (HMM, KF) and one for general graphs (LBP, JTA).
• Moved JTA to supplementary.
• Moved a lot of the material about SSM back from supplementary into main.
• Fixed more typos.
• Updated figure links to point to https://github.com/probml/pyprobml/blob/master/notebooks.md, so we no longer need to auto-generate figure names or use firestore.

2022-06-24

• moved some more material (eg on HMM, LDS) back from supplementary into main
• added more latexified figures
• fixed more typos
• cleaned up mixFA and PPCA sections
• added stuff on discriminative GMMs

2022-06-20

• moved some of the content on MVNs (eg conjugate analysis) back from supplementary into main text
• add latexified versions of some figures (eg in GP chapter)
• fix more typos

2022-06-17

• remove list of figures (LOF) at the start, which wastes 50 pages (now under 1300 :)
• fix some typos

2022-06-11

• update diffusion models chapter
• fixed >300 typos / errors (mostly found by John Fearns)
• other tweaks

2022-05-18

• URLs in the figure captions now refer to the name of the file, such as mc_estimate_pi.ipynb , not https://code.probml.ai/book2/10.1. Clicking on the link will open that file directly in colab without us having to keep backend firebase up to date as things change. As a consequence, book2.pdf has a list of figures which increases the page count by about 30. This is just a temporary measure to faciliate processing in the backend; this will be removed for camera ready version.
• Added TOC, footer, and line numbering to supp2 (previously accidently omitted)
• Made some tweaks to the ADVI / Gaussian VI section, to add place holder figures for future demos that are being written.
• Tweaked the LKJ section.
• Moved online VI to supplement.
• Moved some common distributions (eg Pareto) from supplement to main text, to reduce cross reference issues.
• Fixed some typos.

2022-05-17

• 35 pages shorter, by moving even more content related to SSMs (models, inference, PF) to the online supplement
• changed LDS notation to "east coast" tradition (F/H, not A/C) for consistency with the nonlinear case
• fixed some typos
• added a few more figures
• cleaned up ADF sections

2022-05-02

• This version now supports cross references from the main book to supplementary material.
• Fixes all the broken cross references in the main book. (Supplementary still has broken xrefs.)
• Fixes some other typos.

2022-04-29

Major re-org. There is now 1 less chapter. Chapters after ch 27 have a different number->name mapping.

• Merged the HMM and SSM chapters to save space. Put the extra HMM stuff in supplementary.
• Removed a lot of the material that was duplicated from vol 1 (on probability distributions and basic conjugate analysis) to save space.
• Moved a lot of the background material on probability distributions to supplementary.
• Moved the section on topic modeling back from supplementary into main text.
• Moved some PGM sections back from supplementary into main text.
• Moved the sections on PGM structure learning back from supplementary into main text.
• Moved stuff on derivative free optimization back from supplementary into main text.
  The overall page count is basically the same as before, but there is now fewer holes in the main text.

2022-04-26

• Fixed many typos
• Expanded the discussion of message passing on a junction tree, including Gaussian case
• Tweaked forwards-backwards section
• Added section on sparse Gaussian VI (Barfoot)
• Added Gaussian BP example
• Updated some GP figures