Inputs: | ||
N | - | Number of samples to generate each iteration |
g | - | Number of these samples to use to update parameters |
alpha | - | Smoothing parameter |
k | - | Number of cluster means to find |
data | - | The data we are trying to fit means to (Should be n x d, |
where there are n points, and d dimensions) | ||
modif | - | If 1, use modified smoothing, |
otherwise use standard smoothing | ||
drplot | - | If 1, draws the cluster means |
and the data (for 2-dimensions) | ||
c | - | Optional starting centroids |
sigma0 | - | Optional starting standard deviation |
Outputs: | ||
mu | - | The centroids found via the CE method, using Normal updating |
with the parameter set | ||
count | - | The number of iterations taken |
score | - | The final score of these centroids |
Inputs: | ||
N | - | Number of samples to generate each iteration |
rho | - | The fraction of samples used to update the probabilities |
alpha | - | Smoothing parameter |
k | - | Number of clusters to assign points to |
data | - | The data we are trying to assign to clusters (Should be n x d, |
where there are n points, and d dimensions) | ||
Outputs: | ||
x | - | The best found assignment of the data points |
Inputs: | ||
N | - | Number of samples to generate each iteration |
rho | - | The fraction of samples used to update the probabilities |
alpha | - | Smoothing parameter |
k | - | Number of clusters to assign points to |
data | - | The data we are trying to assign to clusters (Should be n x d, |
where there are n points, and d dimensions) | ||
Outputs: | ||
x | - | The best found assignment of the data points |
Inputs: | ||
c | - | A set of cluster means |
data | - | The data we are trying to assign to clusters (Should be n x d, |
where there are n points, and d dimensions) | ||
k | - | Number of clusters to assign points to |
Outputs: | ||
x | - | The assignment of the data points to clusters |
Inputs: | ||
x | - | A labelling of data points |
data | - | The data we are trying to fit means to (Should be n x d, |
where there are n points, and d dimensions) | ||
k | - | Number of cluster means |
Outputs: | ||
x | - | The cluster means calculated for this labelling of data points |
imim.mat - this is the image data file : it contains 2 variables, I : the 20 by 20 subimage (ie the actual image), Ib: the 256 by 25 set of all 5x5 subimages of I.
NCEICJ.m - nce with component upd. and injection
clim.m - create all pxp subimages from a given image (ie to generate Ib, can use this)
mu2im.m - converts the K cluster means back into pxp matrices so its easier to plot.