PLOT_EXTRA plots data on the GUI specific to the dataset type
Description:
Function to plot data on the GUI specific to the dataset type. For
example, for Classification, the Classification Error is plotted.
Syntax:
PLOT_EXTRA(TrainingParams,tiOrder,tiAll,TrainingResults);
PLOT_EXTRA(TrainingParams,tiOrder,tiAll,TrainingResults,tiAllTest,TestingResults);
Input Parameters:
o TrainingParams: A structure that contains information for training a
theta neuron network, such as the learning method. This structure is
generated by get_training_params.
o tiOrder: The order in which the input patterns were processed during this
epoch. The order is random when using online learning.
o tiAll: A cell array of length a, the number of input patterns, of which
each cell is a qx2 array that contains the neuron indices for each
spike time and the input spike times. q may vary from cell to cell.
o TrainingResults: A structure containing error gradient with respect to
the weights (DEDW) and delays (DEDTau). The structure also includes
the SSE, NonFireFlag to indicate is any neurons are not firing along
with NonFireCount, an array used to keep track of how many times a
neuron has not fired over multiple input patterns. This structure is
generated by calculate_gradient. If passed as an input, certain
results are appended.
o tiAllTest: A cell array of input test patterns in the same format as
tiAll.
o TestingResults: A structure containing results generated on the testing
data in the same format as TrainingResults.
Output Parameters:
o None
Example:
>> %Assumes valid simulation training results are loaded below
>> GUIHandle=start_ThNN;
>> Temp=get(GUIHandle,'UserData');
>> Handles=Temp{2};
>> cd('Results');
>> [File,PathLoad] = uigetfile('*.mat','Load Simulation...');
>> cd('..');
>> load([PathLoad, File]);
>> TrainingParams.Handles=Handles;
>> set_params(Handles,ThNN,TrainingParams);
>> plot_extra(TrainingParams,tiOrder,tiAll,TrainingResults);
See also theta_neuron_network0001 function plot_extra(TrainingParams,tiOrder,tiAll,TrainingResults,tiAllTest,TestingResults) 0002 %PLOT_EXTRA plots data on the GUI specific to the dataset type 0003 % 0004 %Description: 0005 %Function to plot data on the GUI specific to the dataset type. For 0006 %example, for Classification, the Classification Error is plotted. 0007 % 0008 %Syntax: 0009 %PLOT_EXTRA(TrainingParams,tiOrder,tiAll,TrainingResults); 0010 %PLOT_EXTRA(TrainingParams,tiOrder,tiAll,TrainingResults,tiAllTest,TestingResults); 0011 % 0012 %Input Parameters: 0013 %o TrainingParams: A structure that contains information for training a 0014 % theta neuron network, such as the learning method. This structure is 0015 % generated by get_training_params. 0016 %o tiOrder: The order in which the input patterns were processed during this 0017 % epoch. The order is random when using online learning. 0018 %o tiAll: A cell array of length a, the number of input patterns, of which 0019 % each cell is a qx2 array that contains the neuron indices for each 0020 % spike time and the input spike times. q may vary from cell to cell. 0021 %o TrainingResults: A structure containing error gradient with respect to 0022 % the weights (DEDW) and delays (DEDTau). The structure also includes 0023 % the SSE, NonFireFlag to indicate is any neurons are not firing along 0024 % with NonFireCount, an array used to keep track of how many times a 0025 % neuron has not fired over multiple input patterns. This structure is 0026 % generated by calculate_gradient. If passed as an input, certain 0027 % results are appended. 0028 %o tiAllTest: A cell array of input test patterns in the same format as 0029 % tiAll. 0030 %o TestingResults: A structure containing results generated on the testing 0031 % data in the same format as TrainingResults. 0032 % 0033 %Output Parameters: 0034 %o None 0035 % 0036 %Example: 0037 %>> %Assumes valid simulation training results are loaded below 0038 %>> GUIHandle=start_ThNN; 0039 %>> Temp=get(GUIHandle,'UserData'); 0040 %>> Handles=Temp{2}; 0041 %>> cd('Results'); 0042 %>> [File,PathLoad] = uigetfile('*.mat','Load Simulation...'); 0043 %>> cd('..'); 0044 %>> load([PathLoad, File]); 0045 %>> TrainingParams.Handles=Handles; 0046 %>> set_params(Handles,ThNN,TrainingParams); 0047 %>> plot_extra(TrainingParams,tiOrder,tiAll,TrainingResults); 0048 % 0049 %See also theta_neuron_network 0050 0051 %Copyright (C) 2008 Sam McKennoch <Samuel.McKennoch@loria.fr> 0052 0053 0054 if nargin==4 0055 TestFlag=0; 0056 else 0057 TestFlag=1; 0058 end 0059 0060 if strcmp(TrainingParams.Type,'Classification') 0061 if TestFlag 0062 plot(TrainingParams.Handles.axes1,[0 1 TrainingParams.TestingFrequency*(1:(length(TrainingResults.ClassErr)-2))],TrainingResults.ClassErr,... 0063 [0 1 TrainingParams.TestingFrequency*(1:(length(TestingResults.ClassErr)-2))],TestingResults.ClassErr); 0064 legend(TrainingParams.Handles.axes1,'% Training Classification Error','% Testing Classification Error','Location','NorthWest'); 0065 else 0066 plot(TrainingParams.Handles.axes1,[0 1 TrainingParams.TestingFrequency*(1:(length(TrainingResults.ClassErr)-2))],TrainingResults.ClassErr);%,... 0067 legend(TrainingParams.Handles.axes1,'% Training Classification Error','Location','NorthWest'); 0068 end 0069 0070 0071 elseif strcmp(TrainingParams.Type,'Regression') %&& testing_valid 0072 Coding=TrainingParams.Coding; 0073 InputNeurons=[]; 0074 OutputNeurons=[]; 0075 %Note: I guess this could cause problems if DEDW goes to exactly zero 0076 %for some connection, in which we could report extra input or output 0077 %neurons. A better solution is needed eventually. 0078 for j=1:size(TrainingResults.DEDW,1) 0079 if sum(TrainingResults.DEDW(:,j)~=0)==0 0080 InputNeurons=[InputNeurons; j]; 0081 end 0082 if sum(TrainingResults.DEDW(j,:)~=0)==0 0083 OutputNeurons=[OutputNeurons; j]; 0084 end 0085 end 0086 0087 Inputs=code_data(InputNeurons, 'DecodeInputs', tiAll, Coding); 0088 Inputs=Inputs(tiOrder); 0089 Outputs=code_data(OutputNeurons, 'EncodeOutputs', TrainingResults.tsCurrent, Coding); 0090 if TestFlag 0091 TestingInputs=code_data(InputNeurons, 'DecodeInputs', tiAllTest, Coding); 0092 TestingOutputs=code_data(OutputNeurons, 'EncodeOutputs', TestingResults.tsCurrent, Coding); 0093 end 0094 0095 if size(Inputs,2)>1 0096 figure; 0097 [X,Y]=meshgrid(Coding.InputRange(1):0.1:Coding.InputRange(2),Coding.InputRange(1):0.1:Coding.InputRange(2)); 0098 RegressionOutput=TrainingParams.FunctionHandle(X,Y); 0099 surf(X,Y,RegressionOutput); 0100 hold on; 0101 if TestFlag 0102 plot3(TrainingParams.Handles.axes1,Inputs(:,1),Inputs(:,2),Outputs,'o',TestingInputs(:,1),TestingInputs(:,2),TestingOutputs,'x'); 0103 legend(TrainingParams.Handles.axes1,'ThNN Train','ThNN Test','Location','North'); 0104 else 0105 plot3(TrainingParams.Handles.axes1,Inputs(:,1),Inputs(:,2),Outputs,'o'); 0106 legend(TrainingParams.Handles.axes1,'ThNN Train','Location','North'); 0107 end 0108 else 0109 RegressionOutput=TrainingParams.FunctionHandle(Coding.InputRange(1):0.1:Coding.InputRange(2)); 0110 if TestFlag 0111 plot(TrainingParams.Handles.axes1,Inputs,Outputs,'o',TestingInputs,TestingOutputs,'x',Coding.InputRange(1):0.1:Coding.InputRange(2),RegressionOutput); 0112 legend(TrainingParams.Handles.axes1,'ThNN Train','ThNN Test','Actual','Location','North'); 0113 else 0114 plot(TrainingParams.Handles.axes1,Inputs,Outputs,'o',Coding.InputRange(1):0.1:Coding.InputRange(2),RegressionOutput); 0115 legend(TrainingParams.Handles.axes1,'ThNN Train','Actual','Location','North'); 0116 end 0117 set(TrainingParams.Handles.axes1,'YLim',[min(RegressionOutput) max(RegressionOutput)]); 0118 end 0119 0120 end 0121 0122 grid(TrainingParams.Handles.axes1,'on'); 0123 pause(0.1);