| |
TOY2D toy data generation object
|
A=TOY(H) returns a toy object initialized with hyperparameters H.
This generates 2d toy data problem of two Gaussians or from a 2d image.
Hyperparameters, and their defaults
l=20 -- examples
seed=-1 -- random seed used to generate it, if -1 do not
set seed
dist -- name of distribution. can be the string: 'gaussians' [default]
or must specify the name of a 'jpg' file (without extension) containing the colors
red = class 1 , blue = class 2 , black
Model
w -- stores label model
Methods:
generate,train,test
Example:
Possible distributions: gaussians,2circles,chess,rectangle_uneven,simple,uneven_gauss
[r a]=train(svm('C=10'),toy2d); plot(a);
pause
clf,[r a]=train(svm({kernel('rbf',0.5),'C=10000'}),toy2d('2circles','l=200')); plot(a);