search

probml / ssm-book

Interactive textbook on state-space models
MIT License
176 stars 14 forks source link

Just curious...is this continuing? #1

Open PaulScemama opened 1 year ago

PaulScemama commented 1 year ago

Hi! I've just started using this interactive book to learn about State Space models for the first time. I only have experience with Bayesian statistics and standard supervised learning (pretty much nothing with any temporal aspect, and minimal experience with dynamical systems), and I've really enjoyed the material. I am at the State Estimation section under the parent section State Space Models.

I've noticed soon I will be getting to sections that have not yet been written. I was wondering if you are planning / currently-working on it? No worries if not, but I think it would be a valuable resource nonetheless.

I've ordered the Bayesian filtering and smoothing book as well to dig deeper into the subject.

Lastly, if you have any other resources that you feel are valuable in studying stochastic dynamical models, I'll take all recommendations! Thanks.

murphyk commented 1 year ago

Hi. Unfortunately this project has been put on ice (although maybe @slinderman will pick it up at some point?). In the meantime, please see chapters 8, 9 and 29 of probml.github.io/book2, and the tutorials at https://probml.github.io/dynamax/

PaulScemama commented 1 year ago

No worries, thanks for letting me know. I'll check out those resources as well.

PaulScemama commented 1 year ago

A quick question @murphyk : in this introduction here you say

"An SSM is a partially observed Markov model."

However, in this article you also wrote here you say that graphical models which have directed edges are called bayesian networks while graphical models with undirected edges are called markov models.

Doesn't an SSM have directed arcs? Wouldn't it be a form of a Bayesian network?

Thanks in advance!