/prog/ - Machine Learning

Name: Anonymous 2013-09-30 16:13

So, anyone done anything to do with Machine Learning?

Name: Anonymous 2013-09-30 19:49

My machine is pretty dumb, I don't think it could learn anything.

Name: Anonymous 2013-09-30 21:22

a little =) still a bit of a ml noob tho =D you?

Name: Anonymous 2013-09-30 22:31

bah no answer yet..
anyway i think centroids are kind of interesting, just mucking about with a few variations.. Split centers / Floating centers / secondary tuning (distance bias / scaling)

Name: Anonymous 2013-09-30 23:23

God shows his love for us in that while we were still sinners, Christ died for us.

Name: Anonymous 2013-09-30 23:32

Yeah.
Assignments.
I hate AI stuff though

Name: Anonymous 2013-09-30 23:52

How about take a human brain, write a function that loops through every sector of the brain, and copies every aspect of every sector's functionality, and then use that automatically-generated "code of the human brain," and the learning processes therein, as your starting point. Then modify that model slightly, to improve upon man's shortcomings. There you go: artificial intelligence.

Name: Anonymous 2013-10-01 1:23

/prog/ challenge: Create a genius-level artificial intelligence using only PHP and a box of matches.

Name: Anonymous 2013-10-01 2:03

>write a function that loops through every sector of the brain
Laughing Neuroscientists.jpg

Name: Anonymous 2013-10-01 5:38

I'm writing a genetic algorithm based system to determine the optimal configuration for a neural net that will decide the best values for my fuzzy logic control that controls my expert system.

Name: Anonymous 2013-10-01 5:43

██████████████████████
      ██████████████████ ████ ██████████████ ████ ██ ██ ██████ ████ ██ ██
    ██ ██ ████ ████████████████████ ██ ██ ██ ██ ██ ██ ██ ██ ██
    ██ ████ ██ ████ ██████████████ ██ ████████ ████ ██ ████
      ██████████████████ ██████ ██ ██████████████ ██ ██ ██ ██ ██ ██ ██ ██ ██
      ██ ████ ██ ██ ██████████████ ████ ██ ██ ██████ ████ ██ ██
      ██████████ ██ ██ ██████████████
      ██ ████ ██ ██ ██████████████ ██████████ ██ ██ ██████ ██ ██
      ██████████ ██ ██ ██████████████ ██ ██ ██ ██ ████ ████
      ██ ████ ██ ██ ██████████████ ██ ████████ ████ ██ ██ ██
      ████████ ██ ██ ██████████████ ██ ██ ██ ██ ██ ██
      ██ ████ ██ ██████████████ ██ ██ ██ ██████ ██ ██
      ████████ ████████ ██████████████
      ████████████ ████████████████████
      ██████████████

Name: Anonymous 2013-10-01 10:38

nice kite >>11

wonder if you can train centroids with an rbf autoencoder

Name: Anonymous 2013-10-01 10:44

vbf-rbm's? ^^

Name: Anonymous 2013-10-01 10:47

here's the vbf bit =D





function [d, dfloat] = getDistxV(mat, cent)



m = size(mat, 1);



n = size(cent,1);



w = size(cent,2);



d(1:m, 1:n) = 0;



dfloat(1:m, 1:n) = 0;



for(iter=1:n);



  cx = repmat(cent(iter,:), m, 1);



  m2 = mat .- cx;



  d2 = sum(m2, 2) ./ w;



  m3 = m2 .- d2(:, ones(1,size(mat,2)));



  d3 = sum(abs(m3), 2);



  d(:,iter) = d3;



  dfloat(:,iter) = d2;



  endfor;



endfunction;

Machine Learning

1 Name: Anonymous 2013-09-30 16:13

2 Name: Anonymous 2013-09-30 19:49

3 Name: Anonymous 2013-09-30 21:22

4 Name: Anonymous 2013-09-30 22:31

5 Name: Anonymous 2013-09-30 23:23

6 Name: Anonymous 2013-09-30 23:32

7 Name: Anonymous 2013-09-30 23:52

8 Name: Anonymous 2013-10-01 1:23

9 Name: Anonymous 2013-10-01 2:03

10 Name: Anonymous 2013-10-01 5:38

11 Name: Anonymous 2013-10-01 5:43

12 Name: Anonymous 2013-10-01 10:38

13 Name: Anonymous 2013-10-01 10:44

14 Name: Anonymous 2013-10-01 10:47