forked from cvjena/knfst
-
Notifications
You must be signed in to change notification settings - Fork 0
/
learn_oneClassNovelty_knfst.m
38 lines (33 loc) · 1.34 KB
/
learn_oneClassNovelty_knfst.m
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
% Learning method for one-class classification with KNFST according to the work:
%
% Paul Bodesheim and Alexander Freytag and Erik Rodner and Michael Kemmler and Joachim Denzler:
% "Kernel Null Space Methods for Novelty Detection".
% Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2013.
%
% Please cite that paper if you are using this code!
%
%
% function model = learn_oneClassNovelty_knfst(K)
%
% compute one-class KNFST model by separating target data from origin in feature space
%
% INPUT:
% K -- (n x n) kernel matrix containing pairwise similarities between n training samples
%
% OUTPUT:
% model -- the one-class KNFST model used in test_oneClassNovelty_knfst
%
% (LGPL) copyright by Paul Bodesheim and Alexander Freytag and Erik Rodner and Michael Kemmler and Joachim Denzler
%
function model = learn_oneClassNovelty_knfst(K)
% get number of training samples
n = size(K,1);
% include dot products of training samples and the origin in feature space (these dot products are always zero!)
K = [K, zeros(n,1); zeros(1,n), 0];
% create one-class labels + a different label for the origin
labels = [ ones(n,1) ; 0 ];
% get model parameters
model.proj = calculateKNFST(K,labels);
model.targetValue = mean(K(labels==1,:)*model.proj);
model.proj = model.proj(1:n,:);
end