[mirtoolbox] Re: Some issues with mircluster
- From: Olivier Lartillot <olartillot@xxxxxxxxx>
- To: mirtoolbox@xxxxxxxxxxxxx
- Date: Wed, 10 Nov 2010 15:25:57 +0200
Hi once again,
Yes, it seems that since MIRtoolbox 1.2, it is not possible anymore to call
mircluster with several features (grouped into one cell array). I wonder if
this feature was of any importance (I never noticed its absence since that
time, 2 years ago). But if the need arises, we could think about its
implementation.
(For your second problem, Alex, it is again your issue with the missing
randperm.m file in your Matlab installation.)
Olivier
Alexander Refsum Jensenius kirjoitti 9.11.2010 kello 19.26:
> Hi,
>
> I am trying to do what it says in the manual (1.3) on p164:
>
>>> mircluster(sg, {mirmfcc(sg), mircentroid(sg)})
>
> This gives me the following error:
>
>> Clustering frames...
>> ??? Undefined function or method 'isnan' for input arguments of type 'cell'.
>>
>> Error in ==> kmeans_clusters at 62
>> if nargin < 2 | isempty(n_max) | isnan(n_max), n_max = ceil(sqrt(dlen)); end
>>
>> Error in ==> mircluster at 113
>> [cc, p, err, ind] =
>> kmeans_clusters(va',option.nclust,option.nruns);
>
>
> When I try the second example:
>>> mircluster(cc)
>
> I get this error:
>
>> Clustering frames...
>> ??? Undefined function or method 'randperm' for input arguments of type
>> 'double'.
>>
>> Error in ==> som_kmeans at 45
>> temp = randperm(l);
>>
>> Error in ==> kmeans_clusters at 94
>> [c, k, err] = som_kmeans('batch', D, i, 100, 0);
>>
>> Error in ==> mircluster at 113
>> [cc, p, err, ind] =
>> kmeans_clusters(va',option.nclust,option.nruns);
>
>
> This is using a single, short wave file as input. What am I doing wrong?
>
> Thanks for any help in clarifying this.
> Alexander
Other related posts:
