luis@199: %% Configuration file used in SMALL_learn
luis@199: %
luis@199: %   Use this file to change the dictionary learning algorithms in SMALLBox
luis@199: %   Please refer to the documentation before editing this file
luis@190: 
luis@199: %   Centre for Digital Music, Queen Mary, University of London.
luis@199: %   This file copyright 2009 Ivan Damnjanovic.
luis@199: %
luis@199: %   This program is free software; you can redistribute it and/or
luis@199: %   modify it under the terms of the GNU General Public License as
luis@199: %   published by the Free Software Foundation; either version 2 of the
luis@199: %   License, or (at your option) any later version.  See the file
luis@199: %   COPYING included with this distribution for more information.
luis@199: %
luis@199: %%
luis@190: 
luis@199: if strcmpi(DL.toolbox,'KSVD')
luis@199:     param=DL.param;
luis@190:     param.data=Problem.b;
luis@199:     
luis@190:     D = eval([DL.name,'(param)']);%, ''t'', 5);']);
luis@199: elseif strcmpi(DL.toolbox,'KSVDS')
luis@199:     param=DL.param;
luis@190:     param.data=Problem.b;
luis@190:     
luis@190:     D = eval([DL.name,'(param, ''t'', 5);']);
luis@199: elseif strcmpi(DL.toolbox,'SPAMS')
luis@190:     
luis@199:     X  = Problem.b;
luis@190:     param=DL.param;
luis@190:     
luis@190:     D = eval([DL.name,'(X, param);']);
luis@190:     %   As some versions of SPAMS does not produce unit norm column
luis@190:     %   dictionaries, we need to make sure that columns are normalised to
luis@190:     %   unit lenght.
luis@190:     
luis@190:     for i = 1: size(D,2)
luis@190:         D(:,i)=D(:,i)/norm(D(:,i));
luis@190:     end
luis@199: elseif strcmpi(DL.toolbox,'SMALL')
luis@190:     
luis@199:     X  = Problem.b;
luis@190:     param=DL.param;
luis@190:     
luis@190:     D = eval([DL.name,'(X, param);']);
luis@190:     %   we need to make sure that columns are normalised to
luis@190:     %   unit lenght.
luis@190:     
luis@190:     for i = 1: size(D,2)
luis@190:         D(:,i)=D(:,i)/norm(D(:,i));
luis@190:     end
luis@190:     
luis@199: elseif strcmpi(DL.toolbox,'TwoStepDL')
luis@199:     
luis@190:     DL=SMALL_two_step_DL(Problem, DL);
luis@190:     
luis@190:     %   we need to make sure that columns are normalised to
luis@190:     %   unit lenght.
luis@190:     
luis@190:     for i = 1: size(DL.D,2)
luis@190:         DL.D(:,i)=DL.D(:,i)/norm(DL.D(:,i));
luis@190:     end
luis@190:     D = DL.D;
luis@190:     
luis@190: elseif strcmpi(DL.toolbox,'MMbox')
luis@199:     
luis@190:     DL = wrapper_mm_DL(Problem, DL);
luis@190:     
luis@190:     %   we need to make sure that columns are normalised to
luis@190:     %   unit lenght.
luis@190:     
luis@190:     for i = 1: size(DL.D,2)
luis@190:         DL.D(:,i)=DL.D(:,i)/norm(DL.D(:,i));
luis@190:     end
luis@199:     D = DL.D;
luis@199:       
luis@190: %   To introduce new dictionary learning technique put the files in
luis@199: %   your Matlab path. Next, unique name <TolboxID> for your toolbox needs
luis@190: %   to be defined and also prefferd API for toolbox functions <Preffered_API>
luis@199: %
luis@190: % elseif strcmpi(DL.toolbox,'<ToolboxID>')
luis@190: %     % This is an example of API that can be used:
luis@190: %     % - get training set from Problem part of structure
luis@190: %     % - assign parameters defined in the main program
luis@190: %
luis@199: %     X  = Problem.b;
luis@190: %     param=DL.param;
luis@190: %
luis@190: %     % - Evaluate the function (DL.name - defined in the main) with
luis@190: %     %   parameters given above
luis@190: %
luis@190: %     D = eval([DL.name,'(<Preffered_API>);']);
luis@190: 
luis@199: else
luis@190:     printf('\nToolbox has not been registered. Please change SMALL_learn file.\n');
luis@190:     return
luis@199: end