Ask HN: Best place to start learning about Markov Chains? • Hacker News Board

Re: Ask HN: Best place to start learning about Markov Chains?

#11

Post by gtycomb » Thu, Apr 11, 2019, 12:47 PM UTC

So many there are. Starting with basic Probability, this lecture series is a good first intro.

https://www.dartmouth.edu/~chance/teaching_aids/books_articl...

Or starting from the basics, and learning how to actually do the number crunching, this is unusually good (Stewart, Introduction to numerical solution of Markov Chains):

https://press.princeton.edu/titles/5640.html

Robert Gallager's MIT lecture series, very well presented, titled Principles of Digital Communications, takes you on another train based on Markov Chains (Kalman filters, etc).

https://ocw.mit.edu/courses/electrical-engineering-and-compu...

Re: Ask HN: Best place to start learning about Markov Chains?

#12

Post by crshults » Thu, Apr 11, 2019, 12:55 PM UTC

I thought this recent post: 'Generating More of My Favorite Aphex Twin Track'[1] had a good beginner-level write up on Markov Chains. [1]https://news.ycombinator.com/item?id=19490832

Re: Ask HN: Best place to start learning about Markov Chains?

#13

Post by dcwca » Thu, Apr 11, 2019, 12:56 PM UTC

Just pick a random place to start, read some stuff, and then take a guess as to which direction to go in next, based on what's probably a good next thing to read. Then keep repeating the process over and over again.

Re: Ask HN: Best place to start learning about Markov Chains?

#14

Post by DanBC » Thu, Apr 11, 2019, 12:58 PM UTC

Not sure it's introductory, but A Mathematical Theory of Communication, page 5 onwards, is useful: http://www.math.harvard.edu/~ctm/home/text/others/shannon/en...

Re: Ask HN: Best place to start learning about Markov Chains?

#15

Post by Scarblac » Thu, Apr 11, 2019, 1:16 PM UTC

dcwca wrote: post #13

Just pick a random place to start, read some stuff, and then take a guess as to which direction to go in next, based on what's probably a good next thing to read. Then keep repeating the process over and over again.

It's also important that you base your guess of what's probably good to read next only on the previous thing you read. Forget everything that came before that.

Re: Ask HN: Best place to start learning about Markov Chains?

#16

Post by pdpi » Thu, Apr 11, 2019, 1:20 PM UTC

dcwca wrote: post #13

Just pick a random place to start, read some stuff, and then take a guess as to which direction to go in next, based on what's probably a good next thing to read. Then keep repeating the process over and over again.

You can then model the probability that you'll end up at any one given place after n steps as a markov chain.

Re: Ask HN: Best place to start learning about Markov Chains?

#17

Post by danaugrs » Thu, Apr 11, 2019, 1:23 PM UTC

I really like this short, relaxed video: "Information Theory part 10: What is a Markov chain?" by Art of the Problem https://www.youtube.com/watch?v=o-jdJxXL_W4

If you like it I recommend watching the whole series.

Re: Ask HN: Best place to start learning about Markov Chains?

#18

Post by i_am_proteus » Thu, Apr 11, 2019, 1:35 PM UTC

If the "motivation-theorem-proof" style appeals to you, find a copy of Finite Markov Chains by Kemeny and Snell. ISBN 0442043287

Re: Ask HN: Best place to start learning about Markov Chains?

#19

Post by zeckalpha » Thu, Apr 11, 2019, 1:49 PM UTC

dcwca wrote: post #13

Just pick a random place to start, read some stuff, and then take a guess as to which direction to go in next, based on what's probably a good next thing to read. Then keep repeating the process over and over again.

Are you describing Markov chains or how to learn about Markov chains?

Re: Ask HN: Best place to start learning about Markov Chains?

#20

Post by lallysingh » Thu, Apr 11, 2019, 1:51 PM UTC

zeckalpha wrote: post #19

dcwca wrote: post #13

Just pick a random place to start, read some stuff, and then take a guess as to which direction to go in next, based on what's probably a good next thing to read. Then keep repeating the process over and over again.

Are you describing Markov chains or how to learn about Markov chains?

It's a meta joke