ACTIVATION Obtain activation maps
A = activation(p,smap,Nd)
INPUTS
p: Data points in the input space
smap: UnioviSOM SOM structure
Nd: Neighborhood distance for lateral interaction (default=1)
OUTPUTS:
A: (1xS) Vector containing the activations for each neuron
DESCRIPTION
This function computes the resulting activation value for each neuron
in the SOM by summing up all the independent activation patterns
originated by each individual stimulus (input data point) in p.
Gaussian lateral interaction with width Nd is considered.
EXAMPLE
p = [randn(3,100)+3 randn(3,100)-3]; % generate two clouds of 3D points
t = [randn(3,500)+3]; % test set similar to the first cloud
som = inisom(p,{30,30},{'x','y','z'}); % initialize a SOM (PCA initialization)
som = bsom(p,som,10,1); % train a SOM (10 epochs, final neigborhood=1)
a = activation(p(:,1:100),som,2); % compute the activation of the first cloud
figure(1);
planes(som); % plot the SOM planes
figure(2);
planes(som.pos,a); % plot the activation map of the test set
title('activation map');