-
Notifications
You must be signed in to change notification settings - Fork 3
/
zftftb_sdi_simscores.m
142 lines (117 loc) · 4.13 KB
/
zftftb_sdi_simscores.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
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
function SCORES=zftftb_sdi_simscores(CONTOUR_GROUP1,CONTOUR_GROUP2,F,T,varargin)
%zftftb_sdi_simscore computes similarity scores between two groups of contours, this
%method is incredibly sensitive, it is recommended to use a bootstrap to determine
%the *expected* difference in similarity scores for sounds from the same *group*
%
% [SCORES]=zftftb_sdi_simscores(CONTOUR_GROUP1,CONTOUR_GROUP2,F,T,varargin)
%
% CONTOUR_GROUP1
% 3D matrix of contours returned by zftftb_sdi (either the re or im field)
%
% CONTOUR_GROUP2
% 3D matrix of contours returned by zftftb_sdi (either the re or im field)
%
% F
% frequency vector returned by zftftb_sdi
%
% T
% time vector returned by zftftb_sdi
%
% the following may be specified as parameter/value pairs:
%
% time_range
% time in seconds to compute the scores over (two element vector, in s,
% e.g. [.2 .8] will compute scores between 200 and 800 ms)
%
% freq_band
% frequency in Hz to compute score over (two element vector, e.g. [2e3 8e3]
% will compute scores between 2 and 8 kHz)
%
% the program returns the following outputs
%
% SCORES
% M X N cell array of similarity scores, where M is the group of contours and N is the SDI
%
% example:
%
% Start by computing the contours for a set of microphone signals with a sampling rate of 24 kHz
%
% [sdi f t contours]=zftftb_sdi(mic_signals,24e3);
%
% Now you want to compare the first 100 trials with the last 100, and compare the signals
% between 200 and 800 msecs and 2000 and 9000 Hz using the imaginary contours
%
% scores=zftftb_sdi_simscores(contours.im(:,:,1:100),contours.im(:,:,end-100:end),f,t,'time_range',[.2 .8],...
% 'freq_range',[2e3 9e3]);
%
% The similarity scores between group 1 contours and SDI 1 (i.e. self-similarity) is scores{1,1}, then
% the cross-similarity between group 2 contours and SDI 1 is scores{2,1}
%
% dprime=(mean(scores{1,1})-mean(scores{2,1}))./std([scores{1,1};scores{2,1}])
%
% This returns a dprime-like measure to compare the difference between the first and second group. Given the sensitivity
% of the measure (0-.4 tends to indicate no effect, >.6 a moderate effect, >1 a strong effect). It's advised to run a
% bootstrap to determine a rigorous cutoff.
nparams=length(varargin);
if mod(nparams,2)>0
error('ephysPipeline:argChk','Parameters must be specified as parameter/value pairs!');
end
time_range=[T(1) T(end)];
freq_band=[3e3 9e3];
for i=1:2:nparams
switch lower(varargin{i})
case 'time_range'
time_range=varargin{i+1};
case 'freq_band'
freq_band=varargin{i+1};
end
end
mint=max(find(T<=time_range(1)));
maxt=min(find(T>=time_range(2)));
minf=max(find(F<=freq_band(1)));
maxf=min(find(F>=freq_band(2)));
if isempty(mint), mint=1; end
if isempty(maxt), maxt=length(T); end
if isempty(minf), minf=1; end
if isempty(maxf), maxf=length(F); end
[rows1,columns1,trials1]=size(CONTOUR_GROUP1);
[rows2,columns2,trials2]=size(CONTOUR_GROUP2);
if (rows1 ~= rows2) | (columns1 ~= columns2)
error('Dimensions are different between two contour groups, cannot proceed...');
end
CONTOUR_GROUP1=CONTOUR_GROUP1(minf:maxf,mint:maxt,:);
CONTOUR_GROUP2=CONTOUR_GROUP2(minf:maxf,mint:maxt,:);
if ~isa(CONTOUR_GROUP1,'double')
CONTOUR_GROUP1=double(CONTOUR_GROUP1);
end
if ~isa(CONTOUR_GROUP2,'double')
CONTOUR_GROUP2=double(CONTOUR_GROUP2);
end
% create new sdi from the contour groups
disp('Forming probability densities...');
[rows1,columns1,trials1]=size(CONTOUR_GROUP1);
sdi1=mean(CONTOUR_GROUP1,3);
sdi2=mean(CONTOUR_GROUP2,3);
norm1=sum(sdi1(:).^2);
norm2=sum(sdi2(:).^2);
% get the self-sim scores
SCORES=cell(2,2);
% first idx is contour group, second is SDI
SCORES{1,1}=zeros(trials1,1);
SCORES{1,2}=zeros(trials1,1);
SCORES{2,1}=zeros(trials2,1);
SCORES{2,2}=zeros(trials2,1);
disp('Computing scores for group 1...');
for i=1:trials1
tmp=CONTOUR_GROUP1(:,:,i);
tmp_norm=sum(tmp(:).^2);
SCORES{1,1}(i)=sum(sum(tmp.*sdi1))./sqrt(norm1*tmp_norm);
SCORES{1,2}(i)=sum(sum(tmp.*sdi2))./sqrt(norm2*tmp_norm);
end
disp('Computing scores for group 2...');
for i=1:trials2
tmp=CONTOUR_GROUP2(:,:,i);
tmp_norm=sum(tmp(:).^2);
SCORES{2,1}(i)=sum(sum(tmp.*sdi1))./sqrt(norm1*tmp_norm);
SCORES{2,2}(i)=sum(sum(tmp.*sdi2))./sqrt(norm2*tmp_norm);
end