Error using ==> assert

Version imprimable

Salut, je débute en matlab: j'ai utilisé une fonction prête en matlab et voici le code de cette fonction

Code:

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
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
 
 
function M = slmetric_pw(X1, X2, mtype, varargin)
%SLMETRIC_PW Compute the metric between column vectors pairwisely
%
% [ Syntax ]
%   - M = slmetric_pw(X1, X2, mtype);
%   - M = slmetric_pw(X1, X2, mtype, ...);
%
% [ Arguments ]
%   - X1, X2:       the sample matrices
%   - mtype:        the string indicating the type of metric
%   - M:            the resulting metric matrix
%
% [ Description ]
%    - M = slmetric_pw(X1, X2, mtype) Computes the metrics between
%      column vectors of X1 and X2 pairwisely, using the metric
%      specified by mtype. 
%
%      Both X1 and X2 are matrices with each column representing a 
%      sample. X1 and X2 should have the same number of rows. Suppose
%      the size of X1 is d x n1, and the size of X2 is d x n2. Then 
%      the output metric matrix M will be of size n1 x n2, in which
%      M(i, j) is the metric value between X1(:,i) and X2(:,j).
%
%    - M = slmetric_pw(X1, X2, mtype, ...) Some metric types requires
%      extra parameters, which should be specified in params.
%
%      The supported metrics of this function are listed as follows: 
%      \{:
%        - eucdist:        Euclidean distance: 
%                          $ ||x - y|| $
%
%        - sqdist:         Square of Euclidean distance: 
%                          $ ||x - y||^2 $
%
%        - dotprod:        Canonical dot product: 
%                          $ <x,y> = x^T * y $ 
%
%        - nrmcorr:        Normalized correlation (cosine angle):
%                          $ (x^T * y ) / (||x|| * ||y||) $
%
%        - corrdist:       Normalized Correlation distance
%                          $ 1 - nrmcorr(x, y) $
%
%        - angle:          Angle between two vectors (in radian)  
%                          $ arccos (nrmcorr(x, y)) $
%        - quadfrm:        Quadratic form:  
%                          $ x^T * Q * y $
%                         Q is specified in the 1st extra parameter 
%
%        - quaddiff:       Quadratic form of difference:
%                          $ (x - y)^T * Q * (x - y) $
%                         Q is specified in the 1st extra parameter 
%
%        - cityblk:        City block distance (abssum of difference)
%                          $ sum_i |x_i - y_i| $
%
%        - maxdiff:        Maximum absolute difference  
%                          $ max_i |x_i - y_i| $
%
%        - mindiff:        Minimum absolute difference
%                          $ min_i |x_i - y_i| $
%
%        - minkowski:      Minkowski distance
%                          $ (\sum_i |x_i - y_i|^p)^(1/p) $
%                         The order p is specified in the 1st extra parameter
%
%        - wsqdist:        Weighted square of Euclidean distance     
%                          $ \sum_i w_i (x_i - y_i)^2 $
%                         the weights w is specified in 1st extra parameter 
%                         as a d x 1 column vector    
%
%        - hamming:        Hamming distance with threshold t
%                          \{
%                              ht1 = x > t
%                              ht2 = y > t
%                              d = sum(ht1 ~= ht2)                  
%                          \}
%                         use threshold t as the first extra param.
%                         (by default, t is set to zero).
%
%        - hamming_nrm:    Normalized hamming distance, which equals the
%                          ratio of the elements that differ.
%                          \{
%                              ht1 = x > t
%                              ht2 = y > t
%                              d = sum(ht1 ~= ht2) / length(ht1)                
%                          \}
%                          use threshold t as the first extra param.
%                         (by default, t is set to zero).
%
%        - intersect:      Histogram Intersection
%                           $ d = sum min(x, y) / min(sum(x), sum(y))$
%
%        - intersectdis:   Histogram intersection distance
%                           $ d = 1 - sum min(x, y) / min(sum(x), sum(y)) $
%
%        - chisq:          Chi-Square Distance
%                           $ d = sum (x(i) - y(i))^2/(2 * (x(i)+y(i))) $
%
%        - kldiv:          Kull-back Leibler divergence
%                           $ d = sum x(i) log (x(i) / y(i)) $
%
%        - jeffrey:        Jeffrey divergence
%                           $ d = KL(h1, (h1+h2)/2) + KL(h2, (h1+h2)/2) $
%      \:}
%
% [ Remarks ]
%   - Both X1 and X2 should be a matrix of numeric values, except
%     for case when metric type is 'hamming' or 'hamming_nrm'. 
%     For hamming or hamming_nrm metric, the input matrix can be logical.
%
% [ Examples ]
%   - Compute different types of metrics in pairwise manner
%     \{
%         % prepare sample matrix
%         X1 = rand(10, 100);
%         X2 = rand(10, 150);
%
%         % compute the euclidean distances (L2) 
%         % between the samples in X1 and X2
%         M = slmetric_pw(X1, X2, 'eucdist');
%
%         % compute the eucidean distances between the samples
%         % in X1 in a pairwise manner
%         M = slmetric_pw(X1, X1, 'eucdist');
%
%         % compute the city block distances (L1)
%         M = slmetric_pw(X1, X2, 'cityblk'); 
%
%         % compute the normalize correlations
%         M = slmetric_pw(X1, X2, 'nrmcorr');
%
%         % compute hamming distances
%         M = slmetric_pw(X1, X2, 'hamming', 0.5);
%         M2 = slmetric_pw((X1 > 0.5), (X2 > 0.5), 'hamming');
%         assert(isequal(M, M2));
%     \}
%
%   - Compute the parameterized metrics
%     \{
%         % compute weighted squared distances with user-supplied weights
%         weights = rand(10, 1);
%         M = slmetric_pw(X1, X2, 'wsqdist', weights);
%
%         % compute quadratic distances (x-y)^T * Q (x-y)
%         Q = rand(10, 10);
%         M = slmetric_pw(X1, X2, 'quaddiff', Q);
%
%         % compute Minkowski distance of order 3         
%         M = slmetric_pw(X1, X2, 'minkowski', 3);
%     \}
%
% [ History ]
%   - Created by Dahua Lin on Dec 06th, 2005
%   - Modified by Dahua Lin on Apr 21st, 2005
%       - regularize the error reporting
%   - Modified by Dahua Lin on Sep 11st, 2005
%       - completely rewrite the core codes based on new mex computation 
%         cores, and the runtime efficiency in both time and space is 
%         significantly increased.
%   - Modified by Dahua Lin on Jul 02, 2007
%       - rewrite the core computation based on the bsxfun introduced in
%         MATLAB R2007a
%       - rewrite the core-mex for cityblk, maxdiff, mindiff
%       - introduce new metrics: corrdist, minkowski
%   - Modified by Dahua Lin on Jul 30, 2007
%       - Add the metric types for histograms, which are originally
%         implemented in slhistmetric_pw in sltoolbox v1.
%   - Modified by Dahua Lin on Aug 16, 2007
%       - revise some of the help contents
%
 
 
%% parse and verify input arguments
error(nargchk(3, inf, nargin));
assert(ischar(mtype), 'sltoolbox:slmetric_pw:invalidarg', ...
    'The metric type should be a string.');
 
if strcmp(mtype, 'hamming') || strcmp(mtype, 'hamming_nrm')
    assert((isnumeric(X1) || islogical(X1)) && ndims(X1) == 2 && ...
           (isnumeric(X2) || islogical(X2)) && ndims(X2) == 2, ...
        'sltoolbox:slmetric_pw:invalidarg', 'X1 and X2 should be numeric or logical matrices.');        
else
    assert(isnumeric(X1) && ndims(X1) == 2 && isnumeric(X2) && ndims(X2) == 2, ...
        'sltoolbox:slmetric_pw:invalidarg', 'X1 and X2 should be numeric matrices.');
end
 
assert(isa(X2, class(X1)), ...
    'sltoolbox:slmetric_pw:invalidarg', 'X1 and X2 should be of the same class.');
 
if isempty(X1) || isempty(X2)
    M = [];
    return;
end
 
 
%% compute
switch mtype        
    case {'eucdist', 'sqdist'}
        checkdim(X1, X2);        
        M = bsxfun(@plus, sum(X1 .* X1, 1)', (-2) * X1' * X2);        
        M = bsxfun(@plus, sum(X2 .* X2, 1), M);        
        M(M < 0) = 0;                        
        if strcmp(mtype, 'eucdist')
            M = sqrt(M);
        end 
 
    case 'dotprod'
        checkdim(X1, X2);        
        M = X1' * X2;
 
    case {'nrmcorr', 'corrdist', 'angle'}
        checkdim(X1, X2);
        ns1 = sqrt(sum(X1 .* X1, 1));
        ns2 = sqrt(sum(X2 .* X2, 1));
        ns1(ns1 == 0) = 1;  
        ns2(ns2 == 0) = 1;
        M = bsxfun(@times, X1' * X2, 1 ./ ns1');
        M = bsxfun(@times, M, 1 ./ ns2);
        switch mtype
            case 'corrdist'
                M = 1 - M;
            case 'angle'
                M = real(acos(M));
        end
 
    case 'quadfrm'
        Q = varargin{1};       
        M = X1' * Q * X2;
 
    case 'quaddiff'
        checkdim(X1, X2);        
        Q = varargin{1};
        M = X1' * (-(Q + Q')) * X2;
        M = bsxfun(@plus, M, sum(X1 .* (Q * X1), 1)');
        M = bsxfun(@plus, M, sum(X2 .* (Q * X2), 1));        
 
    case 'cityblk'
        checkdim(X1, X2);  
        M = pwmetrics_cimp(X1, X2, int32(1));
 
    case 'maxdiff'
        checkdim(X1, X2); 
        M = pwmetrics_cimp(X1, X2, int32(3));
 
    case 'mindiff'
        checkdim(X1, X2);  
        M = pwmetrics_cimp(X1, X2, int32(2));
 
    case 'minkowski'
        checkdim(X1, X2);
        pord = varargin{1};
        if ~isscalar(pord)
            error('sltoolbox:slmetric_pw:invalidparam', ...
                'the mikowski order should be a scalar');
        end
        pord = cast(pord, class(X1));        
        M = pwmetrics_cimp(X1, X2, int32(4), pord);
 
    case 'wsqdist'
        d = checkdim(X1, X2);
        w = varargin{1};
        if ~isequal(size(w), [d, 1])
            error('sltoolbox:slmetric_pw:invalidparam', ...
                'the weights should be given as a d x 1 vector.');
        end              
        wX2 = bsxfun(@times, X2, w);
        M = bsxfun(@plus, (-2) * X1' * wX2, sum(wX2 .* X2, 1));
        clear wX2;        
        wX1 = bsxfun(@times, X1, w);
        M = bsxfun(@plus, M, sum(wX1 .* X1, 1)');      
 
    case {'hamming', 'hamming_nrm'}
        checkdim(X1, X2);
        if islogical(X1) && islogical(X2)
            H1 = X1;
            H2 = X2;
        else
            if isempty(varargin)
                t = 0;
            else
                t = varargin{1};
                assert(isnumeric(t) && isscalar(t), ...
                    'sltoolbox:slmetric_pw:invalidparam', 't should be a numeric scalar.');
            end
            H1 = X1 > t;
            H2 = X2 > t;
        end
        M = pwhamming_cimp(H1, H2);
        if strcmp(mtype, 'hamming_nrm')
            M = M / size(H1, 1);
        end
 
    case 'intersect'
        checkdim(X1, X2);
        M = pwmetrics_cimp(X1, X2, int32(5));
 
    case 'intersectdis'
        checkdim(X1, X2);
        M = 1 - pwmetrics_cimp(X1, X2, int32(5));
 
    case 'chisq'
        checkdim(X1, X2);
        M = pwmetrics_cimp(X1, X2, int32(6));
 
    case 'kldiv'
        checkdim(X1, X2);
        M = pwmetrics_cimp(X1, X2, int32(7));
 
    case 'jeffrey'
        checkdim(X1, X2);
        M = pwmetrics_cimp(X1, X2, int32(8));
 
    otherwise
        error('sltoolbox:slmetric_pw:unknowntype', 'Unknown metric type %s', mtype);
 
 
end
 
%% Auxiliary function
 
function d = checkdim(X1, X2)
 
d = size(X1, 1);
if d ~= size(X2, 1)
    error('sltoolbox:slmetric_pw:sizmismatch', ...
        'X1 and X2 have different sample dimensions');
end

Mais l’hors de l’appel de cette fonction

Code:

1
2
3
4
5
 
X1 = rand(10, 100);
X2 = rand(10, 150);
weights = rand(10, 1);
M = slmetric_pw(X1, X2, 'wsqdist', weights);

Il me donne cette erreur que je n’est pas compris pour le corrigé.

Code:

1
2
3
4
5
 
??? Error using ==> assert
 
Error in ==> slmetric_pw at 176
assert(ischar(mtype), 'sltoolbox:slmetric_pw:invalidarg', ...

Y-a-t-il quelqu’un S-V-P qui peut m’aider ?

28/10/2009, 15h28
Jerome Briot

Quelle est ta version de MATLAB ?

Que retourne ceci ?

Code:

version

Et que retourne ceci ?

Code:

which assert -all

Voila ce que me retourne version

Code:

1
2
3
4
5
6
7
 
 
version
 
ans =
 
7.0.0.19920 (R14)

Code:

1
2
3
4
5
 
 
which assert -all
 
C:\Documents and Settings\Slaima\Bureau\Dossier\assert.m

Déjà, j’ai le code de la fonction assert.m dans mon dossier de travaille et voila sa code :

Code:

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
 
function assert(varargin)
%assert(testCondition)
%   Evaluates the test condition in the caller's context.  If the result is
%   false, an error is thrown indicating that the test failed.
%   TESTCONDITION should be a string.
%
%assert(truthValue)
%assert(truthValue,errMsg)
%   If TRUTHVALUE is zero, an error is thrown.  The text of the error
%   contains ERRMSG if supplied.  This version of ASSERT was made to
%   promote compatibility with Kevin Murphy's Bayes Net Toolbox (i.e. he
%   uses this syntax).  TRUTHVALUE must be a logical or a number.
%
%EXAMPLES:
%   assert('1==2')      ==> assertion error generated
%   assert(1==2)        ==> assertion error generated (BNT syntax)
%   assert 1==2;        ==> assertion error generated
%   x=1; assert('x==1') ==> no error
%   x=1; assert x==1;   ==> no error
%   assert('{a==3')     ==> error generated by Matlab (bad assert code)
%
%by Gerald Dalley
 
% Make ourselves compatible with Kevin Murphy's Bayes Net toolbox.
if (ismember(nargin,[1 2]) && ...
    (islogical(varargin{1}) || isnumeric(varargin{1})))
  if nargin<2, msg = ''; else, msg = varargin{2}; end
  if ~varargin{1}
    error('assertion violated: %s', msg);
  end
  return
end
 
% Standard string-based usage
for i=1:nargin
    if (~ischar(varargin{i}))
        error('Type HELP ASSERT for usage');
    end
end
 
testCondition = join(' ', varargin);
if (~evalin('caller', testCondition))
    error(['ASSERT FAILED: ' testCondition]);
end
 
 
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function s = join(d,varargin)
% This JOIN function is an inlined version of Gerald Dalley's one posted at the
% Matlab Central website.  It is placed here as a convenience to users that
% have not downloaded it.
 
if (isempty(varargin)), 
    s = '';
else
    if (iscell(varargin{1}))
        s = join(d, varargin{1}{:});
    else
        s = varargin{1};
    end
 
    for ss = 2:length(varargin)
        s = [s d join(d, varargin{ss})];
    end
end
 
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function test
% Copy-and-paste the below code into the command window to test this
% function. 
 
clear all;
 
worked = 0; try, assert('1==2'); catch, worked = 1; end
if (~worked), error('Test failed'); end
 
worked = 1; try, x=1; assert('x==1'); catch, worked = 0; end
if (~worked), error('Test failed'); end
 
worked = 0; try, assert(1==2); catch, worked = 1; end
if (~worked), error('Test failed'); end
 
worked = 0; try, assert('{a==3'); catch, worked = 1; end
if (~worked), error('Test failed'); end

28/10/2009, 16h00
Jerome Briot
OK tout s'explique... la fonction ASSERT que tu utilises n'est pas celle de MATLAB (elle n'est d'ailleurs pas disponible dans ta version).

Le code ne fonctionne pas avec la fonction ASSERT que tu as récupérée je ne sais où...

En même temps, ASSERT sert juste à gérer les erreurs.

Tu peux par exemple remplacer :
Code:

1 2 3 assert(ischar(mtype), 'sltoolbox:slmetric_pw:invalidarg', ... 'The metric type should be a string.');
par
Code:

1 2 3 if ~ischar(mtype) error('The metric type should be a string.'); end
Et faire de même avec tous les autres appels à ASSERT
28/10/2009, 16h03
Jerome Briot
Ou encore plus simple, il suffit de supprimer le second argument de tous les appels à ASSERT dans ton code :
Code:

1 2 assert(ischar(mtype), 'sltoolbox:slmetric_pw:invalidarg', ... 'The metric type should be a string.');
devient simplement :

Code:

assert(ischar(mtype), 'The metric type should be a string.');

J’ai supprimé la fonction assert.m de mon dossier de travaille maintenant puisqu’elle n'est pas celle de MATLAB.
Et j’ai corrigé

Code:

1
2
3
 
assert(ischar(mtype), 'sltoolbox:slmetric_pw:invalidarg', ...
    'The metric type should be a string.');

Avec ceci

Code:

1
2
3
4
 
if ~ischar(mtype)
   error('The metric type should be a string.');
end

et voila mon code maitenant

Code:

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
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
 
function M = slmetric_pw(X1, X2, mtype, varargin)
%SLMETRIC_PW Compute the metric between column vectors pairwisely
%
% [ Syntax ]
%   - M = slmetric_pw(X1, X2, mtype);
%   - M = slmetric_pw(X1, X2, mtype, ...);
%
% [ Arguments ]
%   - X1, X2:       the sample matrices
%   - mtype:        the string indicating the type of metric
%   - M:            the resulting metric matrix
%
% [ Description ]
%    - M = slmetric_pw(X1, X2, mtype) Computes the metrics between
%      column vectors of X1 and X2 pairwisely, using the metric
%      specified by mtype. 
%
%      Both X1 and X2 are matrices with each column representing a 
%      sample. X1 and X2 should have the same number of rows. Suppose
%      the size of X1 is d x n1, and the size of X2 is d x n2. Then 
%      the output metric matrix M will be of size n1 x n2, in which
%      M(i, j) is the metric value between X1(:,i) and X2(:,j).
%
%    - M = slmetric_pw(X1, X2, mtype, ...) Some metric types requires
%      extra parameters, which should be specified in params.
%
%      The supported metrics of this function are listed as follows: 
%      \{:
%        - eucdist:        Euclidean distance: 
%                          $ ||x - y|| $
%
%        - sqdist:         Square of Euclidean distance: 
%                          $ ||x - y||^2 $
%
%        - dotprod:        Canonical dot product: 
%                          $ <x,y> = x^T * y $ 
%
%        - nrmcorr:        Normalized correlation (cosine angle):
%                          $ (x^T * y ) / (||x|| * ||y||) $
%
%        - corrdist:       Normalized Correlation distance
%                          $ 1 - nrmcorr(x, y) $
%
%        - angle:          Angle between two vectors (in radian)  
%                          $ arccos (nrmcorr(x, y)) $
%        - quadfrm:        Quadratic form:  
%                          $ x^T * Q * y $
%                         Q is specified in the 1st extra parameter 
%
%        - quaddiff:       Quadratic form of difference:
%                          $ (x - y)^T * Q * (x - y) $
%                         Q is specified in the 1st extra parameter 
%
%        - cityblk:        City block distance (abssum of difference)
%                          $ sum_i |x_i - y_i| $
%
%        - maxdiff:        Maximum absolute difference  
%                          $ max_i |x_i - y_i| $
%
%        - mindiff:        Minimum absolute difference
%                          $ min_i |x_i - y_i| $
%
%        - minkowski:      Minkowski distance
%                          $ (\sum_i |x_i - y_i|^p)^(1/p) $
%                         The order p is specified in the 1st extra parameter
%
%        - wsqdist:        Weighted square of Euclidean distance     
%                          $ \sum_i w_i (x_i - y_i)^2 $
%                         the weights w is specified in 1st extra parameter 
%                         as a d x 1 column vector    
%
%        - hamming:        Hamming distance with threshold t
%                          \{
%                              ht1 = x > t
%                              ht2 = y > t
%                              d = sum(ht1 ~= ht2)                  
%                          \}
%                         use threshold t as the first extra param.
%                         (by default, t is set to zero).
%
%        - hamming_nrm:    Normalized hamming distance, which equals the
%                          ratio of the elements that differ.
%                          \{
%                              ht1 = x > t
%                              ht2 = y > t
%                              d = sum(ht1 ~= ht2) / length(ht1)                
%                          \}
%                          use threshold t as the first extra param.
%                         (by default, t is set to zero).
%
%        - intersect:      Histogram Intersection
%                           $ d = sum min(x, y) / min(sum(x), sum(y))$
%
%        - intersectdis:   Histogram intersection distance
%                           $ d = 1 - sum min(x, y) / min(sum(x), sum(y)) $
%
%        - chisq:          Chi-Square Distance
%                           $ d = sum (x(i) - y(i))^2/(2 * (x(i)+y(i))) $
%
%        - kldiv:          Kull-back Leibler divergence
%                           $ d = sum x(i) log (x(i) / y(i)) $
%
%        - jeffrey:        Jeffrey divergence
%                           $ d = KL(h1, (h1+h2)/2) + KL(h2, (h1+h2)/2) $
%      \:}
%
% [ Remarks ]
%   - Both X1 and X2 should be a matrix of numeric values, except
%     for case when metric type is 'hamming' or 'hamming_nrm'. 
%     For hamming or hamming_nrm metric, the input matrix can be logical.
%
% [ Examples ]
%   - Compute different types of metrics in pairwise manner
%     \{
%         % prepare sample matrix
%         X1 = rand(10, 100);
%         X2 = rand(10, 150);
%
%         % compute the euclidean distances (L2) 
%         % between the samples in X1 and X2
%         M = slmetric_pw(X1, X2, 'eucdist');
%
%         % compute the eucidean distances between the samples
%         % in X1 in a pairwise manner
%         M = slmetric_pw(X1, X1, 'eucdist');
%
%         % compute the city block distances (L1)
%         M = slmetric_pw(X1, X2, 'cityblk'); 
%
%         % compute the normalize correlations
%         M = slmetric_pw(X1, X2, 'nrmcorr');
%
%         % compute hamming distances
%         M = slmetric_pw(X1, X2, 'hamming', 0.5);
%         M2 = slmetric_pw((X1 > 0.5), (X2 > 0.5), 'hamming');
%         assert(isequal(M, M2));
%     \}
%
%   - Compute the parameterized metrics
%     \{
%         % compute weighted squared distances with user-supplied weights
%         weights = rand(10, 1);
%         M = slmetric_pw(X1, X2, 'wsqdist', weights);
%
%         % compute quadratic distances (x-y)^T * Q (x-y)
%         Q = rand(10, 10);
%         M = slmetric_pw(X1, X2, 'quaddiff', Q);
%
%         % compute Minkowski distance of order 3         
%         M = slmetric_pw(X1, X2, 'minkowski', 3);
%     \}
%
% [ History ]
%   - Created by Dahua Lin on Dec 06th, 2005
%   - Modified by Dahua Lin on Apr 21st, 2005
%       - regularize the error reporting
%   - Modified by Dahua Lin on Sep 11st, 2005
%       - completely rewrite the core codes based on new mex computation 
%         cores, and the runtime efficiency in both time and space is 
%         significantly increased.
%   - Modified by Dahua Lin on Jul 02, 2007
%       - rewrite the core computation based on the bsxfun introduced in
%         MATLAB R2007a
%       - rewrite the core-mex for cityblk, maxdiff, mindiff
%       - introduce new metrics: corrdist, minkowski
%   - Modified by Dahua Lin on Jul 30, 2007
%       - Add the metric types for histograms, which are originally
%         implemented in slhistmetric_pw in sltoolbox v1.
%   - Modified by Dahua Lin on Aug 16, 2007
%       - revise some of the help contents
%
 
 
%% parse and verify input arguments
error(nargchk(3, inf, nargin));
if ~ischar(mtype)
   error('The metric type should be a string.');
end
 
if strcmp(mtype, 'hamming') || strcmp(mtype, 'hamming_nrm')
    assert((isnumeric(X1) || islogical(X1)) && ndims(X1) == 2 && ...
           (isnumeric(X2) || islogical(X2)) && ndims(X2) == 2, ...
        'sltoolbox:slmetric_pw:invalidarg', 'X1 and X2 should be numeric or logical matrices.');        
else
    assert(isnumeric(X1) && ndims(X1) == 2 && isnumeric(X2) && ndims(X2) == 2, 'X1 and X2 should be numeric matrices.');
end
 
assert(isa(X2, class(X1)), 'X1 and X2 should be of the same class.');
 
if isempty(X1) || isempty(X2)
    M = [];
    return;
end
 
 
%% compute
switch mtype        
    case {'eucdist', 'sqdist'}
        checkdim(X1, X2);        
        M = bsxfun(@plus, sum(X1 .* X1, 1)', (-2) * X1' * X2);        
        M = bsxfun(@plus, sum(X2 .* X2, 1), M);        
        M(M < 0) = 0;                        
        if strcmp(mtype, 'eucdist')
            M = sqrt(M);
        end 
 
    case 'dotprod'
        checkdim(X1, X2);        
        M = X1' * X2;
 
    case {'nrmcorr', 'corrdist', 'angle'}
        checkdim(X1, X2);
        ns1 = sqrt(sum(X1 .* X1, 1));
        ns2 = sqrt(sum(X2 .* X2, 1));
        ns1(ns1 == 0) = 1;  
        ns2(ns2 == 0) = 1;
        M = bsxfun(@times, X1' * X2, 1 ./ ns1');
        M = bsxfun(@times, M, 1 ./ ns2);
        switch mtype
            case 'corrdist'
                M = 1 - M;
            case 'angle'
                M = real(acos(M));
        end
 
    case 'quadfrm'
        Q = varargin{1};       
        M = X1' * Q * X2;
 
    case 'quaddiff'
        checkdim(X1, X2);        
        Q = varargin{1};
        M = X1' * (-(Q + Q')) * X2;
        M = bsxfun(@plus, M, sum(X1 .* (Q * X1), 1)');
        M = bsxfun(@plus, M, sum(X2 .* (Q * X2), 1));        
 
    case 'cityblk'
        checkdim(X1, X2);  
        M = pwmetrics_cimp(X1, X2, int32(1));
 
    case 'maxdiff'
        checkdim(X1, X2); 
        M = pwmetrics_cimp(X1, X2, int32(3));
 
    case 'mindiff'
        checkdim(X1, X2);  
        M = pwmetrics_cimp(X1, X2, int32(2));
 
    case 'minkowski'
        checkdim(X1, X2);
        pord = varargin{1};
        if ~isscalar(pord)
            error('sltoolbox:slmetric_pw:invalidparam', ...
                'the mikowski order should be a scalar');
        end
        pord = cast(pord, class(X1));        
        M = pwmetrics_cimp(X1, X2, int32(4), pord);
 
    case 'wsqdist'
        d = checkdim(X1, X2);
        w = varargin{1};
        if ~isequal(size(w), [d, 1])
            error('sltoolbox:slmetric_pw:invalidparam', ...
                'the weights should be given as a d x 1 vector.');
        end              
        wX2 = bsxfun(@times, X2, w);
        M = bsxfun(@plus, (-2) * X1' * wX2, sum(wX2 .* X2, 1));
        clear wX2;        
        wX1 = bsxfun(@times, X1, w);
        M = bsxfun(@plus, M, sum(wX1 .* X1, 1)');      
 
    case {'hamming', 'hamming_nrm'}
        checkdim(X1, X2);
        if islogical(X1) && islogical(X2)
            H1 = X1;
            H2 = X2;
        else
            if isempty(varargin)
                t = 0;
            else
                t = varargin{1};
                assert(isnumeric(t) && isscalar(t),'t should be a numeric scalar.');
            end
            H1 = X1 > t;
            H2 = X2 > t;
        end
        M = pwhamming_cimp(H1, H2);
        if strcmp(mtype, 'hamming_nrm')
            M = M / size(H1, 1);
        end
 
    case 'intersect'
        checkdim(X1, X2);
        M = pwmetrics_cimp(X1, X2, int32(5));
 
    case 'intersectdis'
        checkdim(X1, X2);
        M = 1 - pwmetrics_cimp(X1, X2, int32(5));
 
    case 'chisq'
        checkdim(X1, X2);
        M = pwmetrics_cimp(X1, X2, int32(6));
 
    case 'kldiv'
        checkdim(X1, X2);
        M = pwmetrics_cimp(X1, X2, int32(7));
 
    case 'jeffrey'
        checkdim(X1, X2);
        M = pwmetrics_cimp(X1, X2, int32(8));
 
    otherwise
        error('sltoolbox:slmetric_pw:unknowntype', 'Unknown metric type %s', mtype);
 
 
end
 
%% Auxiliary function
 
function d = checkdim(X1, X2)
 
d = size(X1, 1);
if d ~= size(X2, 1)
    error('sltoolbox:slmetric_pw:sizmismatch', ...
        'X1 and X2 have different sample dimensions');
end

Encore il me donne une autre erreur l’hors de la même l’appel

Code:

1
2
3
4
5
6
 
 
??? Undefined command/function 'bsxfun'.
 
Error in ==> slmetric_pw at 266
        wX2 = bsxfun(@times, X2, w);

Y-a-t-il quelqu’un S-V-P qui peut m’aider ?

28/10/2009, 16h41
slaima15

A titre d’indication le code la fonction slmetric_pw.m que je l'utilise a été pris à partir de ce lien

http://www.mathworks.com/matlabcentr...es-and-metrics
28/10/2009, 16h44
Jerome Briot

Citation:

Envoyé par slaima15

J’ai supprimé la fonction assert.m de mon dossier de travaille maintenant puisqu’elle n'est pas celle de MATLAB.
Et j’ai corrigé

Il fallait modifier toutes les lignes contenant la fonction ASSERT

Peu importe, relis bien mon dernier message, il y a une solution plus simple.
Mais il faut que tu récupère à nouveau la fonction ASSERT que tu viens de supprimer :aie:

Bon sinon pour BSXFUN, ça va être un peu plus compliqué...

Citation:

Envoyé par Dut

Bon sinon pour BSXFUN, ça va être un peu plus compliqué...

On va aller au plus simple...

Tu remplaces le bloc suivant :

Code:

1
2
3
4
5
6
7
8
9
10
11
12
    case 'wsqdist'
        d = checkdim(X1, X2);
        w = varargin{1};
        if ~isequal(size(w), [d, 1])
            error('sltoolbox:slmetric_pw:invalidparam', ...
                'the weights should be given as a d x 1 vector.');
        end              
        wX2 = bsxfun(@times, X2, w);
        M = bsxfun(@plus, (-2) * X1' * wX2, sum(wX2 .* X2, 1));
        clear wX2;        
        wX1 = bsxfun(@times, X1, w);
        M = bsxfun(@plus, M, sum(wX1 .* X1, 1)');

par

Code:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
    case 'wsqdist'
        d = checkdim(X1, X2);
        w = varargin{1};
        if ~isequal(size(w), [d, 1])
            error('sltoolbox:slmetric_pw:invalidparam', ...
                'the weights should be given as a d x 1 vector.');
        end              
 
        wX2 = X2.*repmat(w,1,size(X2,2)); 
        temp = (-2) * X1' * wX2;
        temp2 = sum(wX2 .* X2, 1);
        M = temp + repmat(temp2,size(temp,1),1); 
        wX1 = X1.*repmat(w,1,size(X1,2));
        temp = sum(wX1 .* X1, 1)';
        M = M + repmat(temp,1,size(M,2));

Normalement c'est bon...

28/10/2009, 17h42
slaima15
Mais si je veux tester les autres métriques comme 'intersect' ,'chisq', 'maxdiff'
Comme ceci :
Code:

1 2 3 4 X1 = rand(10, 100); X2 = rand(10, 150); M = slmetric_pw(X1, X2,'intersect')
Il me donne cette erreur
Code:

1 2 ??? Undefined command/function 'pwmetrics_cimp'.
Pourtant j’ai récupérer tout le continue du dossier « pwmetric » dans mon dossier de travail.
Et comme j’ai dis tu peux consulter ce dossier avec ce lien pour voir son contenu
http://www.mathworks.com/matlabcentr...es-and-metrics pour que tu puisses connaitre mon problème.
28/10/2009, 22h07
Jerome Briot

1 pièce(s) jointe(s)
Citation:
Envoyé par slaima15

Il me donne cette erreur

Code:

1 2 ??? Undefined command/function 'pwmetrics_cimp'.
Je pense que ta version de MATLAB ne peut pas utiliser les deux fichiers MEX avec l'extension .mexw32

Supprimes ces fichiers (ou renomme les) et copie les fichiers .dll que je te donne dans l'archive zip attachée dans le même répertoire.

Cela fonctionne-t-il ?
29/10/2009, 02h19
paradize3

tu pourrais aussi mettre a jour ta version de maltab? :)
29/10/2009, 06h55
Jerome Briot

Citation:

Envoyé par paradize3

tu pourrais aussi mettre a jour ta version de maltab? :)

C'est pas faux :aie:
29/10/2009, 13h09
slaima15

Oui ça marche avec le .zip que vous m'avez donner, mais possèdes –vous une idée sur un lien qui me donne un code source matlab pour l’utilisation des métriques comme la distance de Hamming, Weighted Chi-Square Distance, Chi-Square Distance, Histogram Intersection Distance… avec un contenu tous écrit avec matlab.J’ai utilisé Google mais j’ai rien trouvé, pouvez –vous m’aider ??