HybridRFM/normalizeData.m at main · EmoryMLIP/HybridRFM · GitHub

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
39
40
41
42
43
44
function Y = normalizeData(Y,numelFeat)
% Y = normalizeData(Y,numelFeat)
%
% normalize mean and standard deviation as proposed in
% https://www.tensorflow.org/api_docs/python/tf/image/per_image_standardization
%
% Input:
%   Y         - features
%   numelFeat - number of features per example.
%
% Output:
%    Y        - normalized features

if nargin==0
    runMinimalExample
    return
end
szY = size(Y);
Y   = reshape(Y,numelFeat,[]);

Y   = Y - mean(Y,1);

s   = std(Y,[],1);
Y   = Y./max(s,1/sqrt(numelFeat));
Y   = reshape(Y,szY);

function runMinimalExample
Y = setupCIFAR10(50);
figure(1);clf;
subplot(1,2,1);
montageArray(Y,10);
axis equal tight
colormap(flipud(colormap('gray')))
colorbar
title('original images');

[Y] = feval(mfilename,Y,32*32*3);

subplot(1,2,2);
montageArray(Y,10);
axis equal tight
colormap(flipud(colormap('gray')))
colorbar
title('normalized images');