4 ms·

Monte Carlo Methods, Stochastic Optimization

134 points by antman 12y ago

dahveed311 12y ago
Just covered Monte Carlo and stochastic topics in MIT 6.00.2x on edX. Cool stuff!
AC__ 12y ago
I bookmarked this for a later date. I am currently taking a Coursera course and one of the projects touched on Monte Carlo methods, they are super cool, I implemented a rudimentary AI that plays tic-tac-toe(well) in like 70 lines of Python.
- jestinjoy1 12y ago
  Which one is that?
  - AC__ 12y ago
    Hey missed this comment, sorry, the course I am referring to is Principles of Computing. Great course.
- chadzawistowski 12y ago
  Do you have the code on GitHub (or elsewhere)? I'd love to read it.
  - AC__ 12y ago
    http://www.codeskulptor.org/#user39_fju5wBozso4CqOw.py http://www.codeskulptor.org/#user39_fju5wBozso4CqOw.py I was blown away that the computer is able to play with just 10 trials. Monte Carlo methods are a very powerful concept for sure.
Dewie 12y ago
This course is from Harvard so it must be good.
- Chronic31 12y ago
  Lol
jrbancel 12y ago
I took this class when I was in graduate school at Harvard. The methods introduced in this course can be used in virtually every domain. I think everyone involved with some sort of data processing should have at least minimal knowledge of Bayesian inference, simulated annealing, data augmentation, sampling, etc...
kgarten 12y ago
are there download links for the videos somewhere? I don't have flash installed and need to travel tomorrow ;)
guru_meditation 12y ago
I heartily recommend the notebooks published in this course as excellent applied reference material to estimation and optimization. I love it how code and coursework are intermingled, reminiscing me of Knuth's Literate Programming [1] My beef with many other courses offered (including Coursera) is that they use Matlab when it's clearly advantageous to use IPython Notebook as a better experimenting environment. For example, Daphne Koeller's PGM course[2] is still in Matlab and no matter what you do the code looks extremely clumsy and hard to read. N.B. I wrote tens of thousands of lines of Matlab code, including GUI programs, but that does not mean it's a good language to use especially in cases like this. [1] http://en.wikipedia.org/wiki/Literate_programming http://en.wikipedia.org/wiki/Literate_programming [2] https://www.coursera.org/course/pgm https://www.coursera.org/course/pgm

134 points by antman 12y ago

Just covered Monte Carlo and stochastic topics in MIT 6.00.2x on edX. Cool stuff!

AC__ 12y ago

I bookmarked this for a later date. I am currently taking a Coursera course and one of the projects touched on Monte Carlo methods, they are super cool, I implemented a rudimentary AI that plays tic-tac-toe(well) in like 70 lines of Python.

jestinjoy1 12y ago
Which one is that?
- AC__ 12y ago
  Hey missed this comment, sorry, the course I am referring to is Principles of Computing. Great course.
chadzawistowski 12y ago
Do you have the code on GitHub (or elsewhere)? I'd love to read it.
- AC__ 12y ago
  http://www.codeskulptor.org/#user39_fju5wBozso4CqOw.py http://www.codeskulptor.org/#user39_fju5wBozso4CqOw.py I was blown away that the computer is able to play with just 10 trials. Monte Carlo methods are a very powerful concept for sure.

Dewie 12y ago

This course is from Harvard so it must be good.

Chronic31 12y ago
Lol

jrbancel 12y ago

I took this class when I was in graduate school at Harvard. The methods introduced in this course can be used in virtually every domain. I think everyone involved with some sort of data processing should have at least minimal knowledge of Bayesian inference, simulated annealing, data augmentation, sampling, etc...

kgarten 12y ago

are there download links for the videos somewhere? I don't have flash installed and need to travel tomorrow ;)

guru_meditation 12y ago

I heartily recommend the notebooks published in this course as excellent applied reference material to estimation and optimization. I love it how code and coursework are intermingled, reminiscing me of Knuth's Literate Programming [1] My beef with many other courses offered (including Coursera) is that they use Matlab when it's clearly advantageous to use IPython Notebook as a better experimenting environment. For example, Daphne Koeller's PGM course[2] is still in Matlab and no matter what you do the code looks extremely clumsy and hard to read. N.B. I wrote tens of thousands of lines of Matlab code, including GUI programs, but that does not mean it's a good language to use especially in cases like this. [1] http://en.wikipedia.org/wiki/Literate_programming http://en.wikipedia.org/wiki/Literate_programming [2] https://www.coursera.org/course/pgm https://www.coursera.org/course/pgm