Visualizing popular machine learning algorithms • Hacker News Board

Re: Visualizing popular machine learning algorithms

#31

Post by andrelaszlo » Sun, Oct 25, 2015, 3:22 PM UTC

There's a bug somewhere.

Refresh, choose dataset: curved, algorithm: k means clustering. You get this:

http://imageshack.com/a/img633/7110/sfteaE.png

If you play around and select different algorithms before selecting k means clustering you can get very different results. :)

Re: Visualizing popular machine learning algorithms

#32

Post by orliesaurus » Sun, Oct 25, 2015, 4:05 PM UTC

andrelaszlo wrote: post #31

There's a bug somewhere. Refresh, choose dataset: curved, algorithm: k means clustering. You get this: http://imageshack.com/a/img633/7110/sfteaE.png If you play around and select different algorithms before selecting k means clustering you can get very different results. :)

some content is loaded over HTTP rather than HTTPS so thas why it might display a blank page for some people who have HTTPS forced

Re: Visualizing popular machine learning algorithms

#33

Post by wybiral » Sun, Oct 25, 2015, 5:41 PM UTC

blt wrote: post #7

For learners it is confusing to see the nonlinear decision boundaries for linear and logistic regression, IMO a note about the feature expansion should be added

Good point, I've updated my post. For linear and logistic regression there's cubic expansion on the features (which is how they can fit curved problems). The relevant Javascript code is on lines 91 and 96.

PS: It can be changed to "linear" or "quadratic" as well.

Re: Visualizing popular machine learning algorithms

#34

Post by wybiral » Sun, Oct 25, 2015, 5:44 PM UTC

andrelaszlo wrote: post #31

There's a bug somewhere. Refresh, choose dataset: curved, algorithm: k means clustering. You get this: http://imageshack.com/a/img633/7110/sfteaE.png If you play around and select different algorithms before selecting k means clustering you can get very different results. :)

I accidentally left k means in there as an option and it doesn't make much sense in the context of this example. So, yeah, it's a bit of a bug. Realistically, linear regression doesn't make sense being included either but it still kinda works.

Re: Visualizing popular machine learning algorithms

#35

Post by wybiral » Sun, Oct 25, 2015, 5:48 PM UTC

0x99 wrote: post #10

Wow it is awesome. Can we also play around with classifier parameters ( k of kNN) ?

Only in the code :) Click "Edit in JSFiddle" and look at line 109 in the Javascript section. You'll see: options.k = 5

Re: Visualizing popular machine learning algorithms

#36

Post by wybiral » Sun, Oct 25, 2015, 5:52 PM UTC

mtw wrote: post #3

Awesome. Would be great to have execution times. Also what is nerdy.js? I saw it was related to "Carl Edward Rasmussen" but couldn't find another reference on the net

It's a Javascript library I put together a long time ago for dealing with datasets and machine learning algorithms. It was used for some of my own personal projects and hasn't been focused on for release in the wild (although I'm considering it now).

The reference to Carl Edward Rasmussen is because I based my minimize function heavily off of this one: http://learning.eng.cam.ac.uk/carl/code/minimize/

Re: Visualizing popular machine learning algorithms

#37

Post by mtw » Sun, Oct 25, 2015, 6:20 PM UTC

wybiral wrote: post #36

mtw wrote: post #3

Awesome. Would be great to have execution times. Also what is nerdy.js? I saw it was related to "Carl Edward Rasmussen" but couldn't find another reference on the net

It's a Javascript library I put together a long time ago for dealing with datasets and machine learning algorithms. It was used for some of my own personal projects and hasn't been focused on for release in the wild (although I'm considering it now). The reference to Carl Edward Rasmussen is because I based my minimize function heavily off of this one: http://learning.eng.cam.ac.uk/carl/code/minimize/

I'd be interested in the library :D

Re: Visualizing popular machine learning algorithms

#38

Post by wybiral » Mon, Oct 26, 2015, 12:31 AM UTC

revorad wrote: post #30

Can someone please explain this?

It's using the X and Y location of the dots as training data. Each algorithm is being trained on (x,y)->color in an attempt to buildup a rule for predicting what color an unseen (x,y) pair would be. The hypothesis it builds is then used to color the background so that you can see the decision boundary.

Re: Visualizing popular machine learning algorithms

#39

Post by treeform » Sat, Nov 07, 2015, 6:30 AM UTC

wybiral wrote: post #36

mtw wrote: post #3

Awesome. Would be great to have execution times. Also what is nerdy.js? I saw it was related to "Carl Edward Rasmussen" but couldn't find another reference on the net

It's a Javascript library I put together a long time ago for dealing with datasets and machine learning algorithms. It was used for some of my own personal projects and hasn't been focused on for release in the wild (although I'm considering it now). The reference to Carl Edward Rasmussen is because I based my minimize function heavily off of this one: http://learning.eng.cam.ac.uk/carl/code/minimize/

Me too!