MATLAB Answers

0

Example from patternnet docs fails in Neural Network Toolbox Version 10.0 (R2017a)

Asked by Terrance Nearey on 16 Sep 2017
Latest activity Edited by Terrance Nearey on 18 Sep 2017
I'd remove this post now if I could. Tracking down the function name conflict has been difficult. The code attached to this answer https://www.mathworks.com/matlabcentral/answers/100197-is-there-any-way-to-list-all-shadowed-files-in-the-matlab-path-in-matlab-7-8-2009a#answer_109545 got me started.
Original problem: Trying to run the example from
help paternnet
but get:
Index exceeds matrix dimensions.
Error in initnw>initialize_layer (line 168)
range(inputStart(j):inputStop(j),:) = temp2((inputStart(j):inputStop(j))-inputStart(j)+1,:);
Complete code (revised):
% NNTest from help patternnet
clear all
clc
dbstop if error
[x,t] = iris_dataset;
net = patternnet(10);
net = train(net,x,t);
Complete output:
Index exceeds matrix dimensions.
Error in initnw>initialize_layer (line 168)
range(inputStart(j):inputStop(j),:) = temp2((inputStart(j):inputStop(j))-inputStart(j)+1,:);
Error in initnw (line 93)
out1 = initialize_layer(in1,in2);
Error in initlay>initialize_network (line 147)
net = feval(initFcn,net,i);
Error in initlay (line 89)
out1 = initialize_network(in1);
Error in network/init (line 32)
net = feval(initFcn,net);
Error in network/configure (line 243)
net = init(net);
Error in nntraining.config (line 116)
net = configure(network(net),X,T);
Error in nntraining.setup>setupPerWorker (line 68)
[net,X,Xi,Ai,T,EW,Q,TS,err] = nntraining.config(net,X,Xi,Ai,T,EW,configNetEnable);
Error in nntraining.setup (line 43)
[net,data,tr,err] = setupPerWorker(net,trainFcn,X,Xi,Ai,T,EW,enableConfigure);
Error in network/train (line 335)
[net,data,tr,err] = nntraining.setup(net,net.trainFcn,X,Xi,Ai,T,EW,enableConfigure,isComposite);
Error in NNTest (line 7)
net = train(net,x,t);
Version information:
MATLAB Version: 9.2.0.556344 (R2017a)
Neural Network Toolbox Version 10.0 (R2017a)
OS info:
Mac OS X 10.11.6 (15G1611)

  2 Comments

The code
[x,t] = iris_dataset;
net = patternnet(10);
net = train(net,x,t);
view(net)
y = net(x);
perf = perform(net,t,y);
classes = vec2ind(y);
works for me in R2017a.
Thanks Walter. Setting path to default via restoredefaultpath fixed everything for me. I'll always try this first before posting issues. I must be forcing some Mathworks function to be shadowed by something my personal utilities, which I usually load at startup.

Sign in to comment.

2 Answers

Answer by Greg Heath
on 17 Sep 2017
 Accepted Answer

clear all
clc
%stoperr
%Undefined function or variable 'stoperr'.
% [x,t] = iris_dataset;
% net = patternnet(10);
% net = train(net,x,t);
% view(net)
% y = net(x);
% perf = perform(net,t,y) %0.0302
% classes = vec2ind(y);
% end
% Error: Illegal use of reserved keyword "end".
close all, clear all, clc
[ x,t ] = iris_dataset;
[ I N ] = size(x)% [ 4 150 ]
[ O N ] = size(t)% [ 3 150]
truclass = vec2ind(t); % 1 to 3
MSEref = mse(t-mean(t,2)) % 0.2222
rng(0), Ntrials = 10
for h = 1:10 % No. of hidden nodes
net = patternnet(h);
for n = 1:Ntrials
net = configure(net,x,t);
[net tr y e ] = train(net,x,t);
% y = net(x); e = t-y;
NMSE(h,n) = mse(e)/MSEref;
predclass = vec2ind(y);
PCTERR(h,n) = 100*mean(predclass~=truclass);
end
end
PCTERR = PCTERR
% Columns 1 through 5
%
% 2.6667 2.0000 2.0000 3.3333 2.6667
% 2.0000 1.3333 2.6667 2.0000 2.0000
% 2.0000 2.0000 4.0000 0.6667 2.6667
% 1.3333 4.0000 4.0000 1.3333 1.3333
% 2.0000 3.3333 3.3333 2.6667 1.3333
% 3.3333 2.0000 2.6667 1.3333 2.0000
% 2.6667 4.6667 3.3333 4.0000 3.3333
% 1.3333 1.3333 2.6667 2.0000 4.6667
% 2.6667 0.6667 2.6667 1.3333 1.3333
% 2.0000 1.3333 1.3333 1.3333 4.0000
%
% Columns 6 through 10
%
% 3.3333 2.6667 2.0000 2.6667 2.0000
% 1.3333 68.6667 1.3333 2.6667 6.0000
% 1.3333 2.6667 2.6667 1.3333 1.3333
% 1.3333 3.3333 3.3333 4.0000 2.6667
% 3.3333 2.0000 1.3333 2.0000 1.3333
% 1.3333 1.3333 1.3333 2.6667 1.3333
% 2.0000 2.6667 2.6667 2.6667 1.3333
% 2.0000 1.3333 1.3333 2.0000 1.3333
% 3.3333 4.0000 3.3333 2.0000 2.0000
% 1.3333 1.3333 2.0000 2.6667 1.3333
Hope this helps,
Thank you for formally accepting my answer
Greg

  2 Comments

P.S.
Any one interested in calculating the correlation coefficient between the 100 samples of NMSE and PCTERR?
Thanks Greg ( and commenter Walter). The source of my problem is almost certainly a function on in my personal utilities library path. Setting executing restoredefaultpath fixes everything. I'll remember to try that first. Now to track down what function is causing the problem. Can't live without my utilities.

Sign in to comment.


Answer by Greg Heath
on 17 Sep 2017
Edited by Greg Heath
on 17 Sep 2017

I get two error messages:
clear all clc
%stoperr
%Undefined function or variable 'stoperr'.
[x,t] = iris_dataset;
net = patternnet(10);
net = train(net,x,t);
view(net)
y = net(x);
perf = perform(net,t,y);
classes = vec2ind(y);
% end
% Error: Illegal use of reserved keyword "end".
Hope this helps,
Greg

  0 Comments

Sign in to comment.