model parameter must be a string error

I have trained a neural network. I am working on a gui which is supposed to get the input from user and use the network to classify it. wen I run my code it executes til sim command and displays model parameter must be a string error..

function varargout = popGUI(varargin)

gui_Singleton = 1;
gui_State = struct('gui_Name',       mfilename, ...
                   'gui_Singleton',  gui_Singleton, ...
                   'gui_OpeningFcn', @popGUI_OpeningFcn, ...
                   'gui_OutputFcn',  @popGUI_OutputFcn, ...
                   'gui_LayoutFcn',  [] , ...
                   'gui_Callback',   []);
if nargin && ischar(varargin{1})
    gui_State.gui_Callback = str2func(varargin{1});
end

if nargout
    [varargout{1:nargout}] = gui_mainfcn(gui_State, varargin{:});
else
    gui_mainfcn(gui_State, varargin{:});
end
% End initialization code - DO NOT EDIT

% --- Executes just before popGUI is made visible.
function popGUI_OpeningFcn(hObject, eventdata, handles, varargin)
global ann

% Choose default command line output for popGUI
handles.output = hObject;

% Update handles structure
guidata(hObject, handles);

% UIWAIT makes popGUI wait for user response (see UIRESUME)
% uiwait(handles.figure1);

% --- Outputs from this function are returned to the command line.
function varargout = popGUI_OutputFcn(hObject, eventdata, handles) 

% Get default command line output from handles structure
varargout{1} = handles.output;

% --- Executes on button press in pushbutton1.
function pushbutton1_Callback(hObject, eventdata, handles)
[A,B]=uigetfile('*.sdf','sdf file');
c=cd;
cd(B);
copyfile(A,'input.sdf')

% --- Executes on button press in pushbutton2.
function pushbutton2_Callback(hObject, eventdata, handles)
global ann
Fig=figure('NumberTitle', 'off');
set(Fig,'Name','My program:Calculating descriptors...');
drawnow;
!mold2.exe input.sdf
D=importdata('descriptors.txt','\t',1)
Mol_no=D.data(:,1);
D.data(:,1)=[]
D.data(:,1)=[]
ain=D.data';
skin=D.data;
handles.ain=ain;
handles.skin=skin;
guidata(hObject,handles)
set(Fig,'Name','My program:Analysing structures...');
drawnow;

popup_sel_index = get(handles.popupmenu1, 'Value');
switch popup_sel_index
    case 1
        netin=handles.ain;
        [netout]=sim(ann,netin)
        A_result=round(netout);
        A_result=A_result';
        A_output=[Mol_no A_result]
        dlmwrite('output.txt',A_output,'delimiter','\t')
    case 2
        S_result=svmclassify(SVMStruct,handles.skin);
        S_output=[Mol_no S_result];
        dlmwrite('output.txt',S_output,'delimiter','\t')
    case 3
        K_output=knnclassify(handles.skin,trainV,trainT);
        dlmwrite('output.txt',K_output,'delimiter','\t')
    case 4
        netin=D.data';
        [netout]=sim(ann,netin)
        A_result=round(netout);
        A_result=A_result';
        S_result=svmclassify(SVMStruct,D.data);
        K_output=knnclassify(D.data,trainV,trainT); 
        ensemblein=[A_result S_result K_result];
        [enout]=sim(ensemble,ensemblein);
        enres=round(enres);
        enres=enres';
        E_output=[Mol_no enres];

end 
guidata(hObjects,handles);

% --- Executes on selection change in popupmenu1.
function popupmenu1_Callback(hObject, eventdata, handles)
%

% --- Executes during object creation, after setting all properties.
function popupmenu1_CreateFcn(hObject, eventdata, handles)

if ispc && isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor'))
    set(hObject,'BackgroundColor','white');
end

set(hObject, 'String', {'ANN', 'SVM', 'KNN', 'Ensemble'});

% --- Executes on button press in pushbutton3.
function pushbutton3_Callback(hObject, eventdata, handles)
command = sprintf ( 'notepad %s','output.txt');
system ( command )

Answer 1

[Ram]

I tried saving my networks(ann and ensemble)in a separate mat file and loaded that into the above mentioned code( instead of declaring them as a global variable) and that worked :)

Answer 2

[Jan]

You forgot to ask a question. There are multiple "sim" commands in your code and we cannot guess, which one causes the error. But you can find this out easily:

dbstop if error

The let your progran run again. Matlab will stop, when the error occurs and you can insspect the values of the variables in the command window.

I know that debugging is a hard job. But of course every programmer has to do it. Debugging is much more efficient than posting here, because your local Matlab knows exactly, where the problem is...

[Ram]

actually all of them cause the same error.. am a beginner as far as programming is concerned and hav jus started to learn.. so wil take sum time to know my way around.. :)

in spite of multiple instances of sim command only one case wil be run depending on user's input so for example if case 1 is run.. it gives the above mentioned error.. actually i have my network(ann) in workspace and i hav declared it as a global variable would that cause any problem??

[Jan]

Ok. Start as the suggested "dbstop if error". Then inspect the parameters, which are not strings, although Matlab wants some. Finally try to assure, that these parameters are strings.

[Ram]

thanks for spending ur valuable time to help me out.. :)

Answer 3

[Rajiv Singh]

This usually happens when you inadvertently call the function SIM (which is /toolbox/simulink/simulink/sim) rather than the method SIM of the Neural Network's "network" object (which is toolbox/nnet/nnet/@network/sim.m). That would happen if the variable you thought represented an object among the input arguments of SIM is something else. Try putting breakpoints at places where invoke the sim method and check the input datatypes (for ann, ensemble).

Rajiv

[Ram]

thank u :)