% File Name : helpinfo.m
% Purpose : help informations
% Author : Hossam E. Mostafa Abdelbaki, School of Computer Science,
% University of Centeral Florida (UCF).
% Release : ver. 1.0.
% Date : October 1998.
%
% RNNSIM is a software program available to the user without any
% license or royalty fees. Permission is hereby granted to use, copy,
% modify, and distribute this software for any purpose. The Author
% and UCF give no warranty, express, implied, or statuary for the
% software including, without limitation, waranty of merchantibility
% and warranty of fitness for a particular purpose. The software
% provided hereunder is on an "as is" basis, and the Author and the
% UCF has no obligation to provide maintenance, support, updates,
% enhancements, or modifications.
%
% RNNSIM is available for any platform (UNIX, PCWIN, MACHITOCH).
% It runs under MATLAB ver. 5.0 or highrer.
%
% User feedback, bugs, or software and manual suggestions can
% be sent via electronic mail to : ahossam@cs.ucf.edu
function info = helpinfo
ind = 0;
ind = ind + 1;
matList(ind).Name='Introduction';
matList(ind).Help= {...
' Artificial Neural Networks '
' =========================== '
' Artificial neural networks are parallel computational models'
'comprised of densely interconnected adaptive processing units '
'A very important feature of these networks is '
'their adaptive nature where learning by example replaces '
'programming in solving problems. This feature makes '
'these computational models very appealing in application '
'domains, where one has little or incomplete understanding '
'of the problem to be solved, but where training data '
'(examples) are available. Another key feature is the '
'intrinsic parallel architecture that allows for fast '
'computation of solutions when these networks are '
'implemented on parallel digital computers or when '
'implemented in customised hardware. Artificial neural '
'networks are viable and very important computational models '
'for a wide variety of problems. These include pattern '
'classification, speech synthesis and recognition, function '
'approximation, image compression, associative memory, '
'clustering, forecasting and prediction, combinatorial '
'optimisation, and non-linear system modeling and control. '
' The networks are neural in the sense that they have '
'been inspired by neuroscience, the study of the human brain '
'and nervous system. The artificial neurons used are thought '
'to be very simple models of their biological counterpart. '
'However, this does not mean that they are faithful models '
'of biological neural or cognitive phenomena, those are of a '
'much more complex nature (Click on the RNN Tutorial button in '
'the main window to see how the Random Neural Network model '
'differs from the other Neural Network models). In fact, the '
'majority of the neural networks presently used are more closely '
'related to traditional mathematical and/or statistical models, '
'such as non-parametric pattern classifiers, non-linear filters '
'and statistical regression models, than they do to neuro '
'biological models. Still, the technology of neural networks '
'attempts to mimic natures approach to solve certain complex '
'problems that are impossible to solve with the more traditional '
'techniques. '
' ====================================================== '
' Random Neural Network '
' =========================== '
' Click on the RNN Tutorial push button in the main program '
'window. '};
ind=ind+1;
matList(ind).Name = ' Typical Problem';
matList(ind).Help= {...
' Typical Problem '
' ================== '
' To illustrate the capability of the Neural Networks (NN )in '
'a real application, we will introduce a certain problem in '
'communications and then we will propose a solution using the NN'
' '
'1- The Problem '
' In Spread Spectrum Communications, the digital data can be '
'encoded using a sequence called m-sequence. This sequence is a '
'cyclic code. To be specific, we will consider m-sequence of '
'length 7 in binary form ( 0 1 0 0 1 1 1) or in its bipolar '
'form (-1 1 -1 -1 1 1 1). This code is decoded using the '
'digital matched filter. A problem appears due to the relative '
'distance between the transmitter and the receiver, this problem'
'is called code acquisition. The transmitter may for example '
'sends (-1 1 -1 -1 1 1 1) and due to the relative distance, the '
'receiver receives it (1 1 1 -1 1 -1 -1). The received code is '
'valid but due to the shift(3 bit) the receiver may deal with '
'it as an invalid code. '
' '
'2- Coventioal Solution '
' To solve this problem the receiver first compares the '
'incoming code with all the possible delayed versions of the '
'basic code and then locally generate the code with the detected'
'delay. This method is called Serial Search Technique and of '
'course it takes a considerable amount of time specially for '
'long sequences. '
' '
'3- Neural Solution '
' We will construct a Random Neural Network (RNN) with 7 input'
'nodes and 7 output nodes. This network will be used (after '
'trained) as a code delay recognizer. Each of the input nodes '
'will be assigned to a bit of the incoming code and the output '
'of each output node will indicate a certain delay. To do so we '
'should prepare some training patterns and target patterns and '
'apply them to the RNN. '
' '
'*** The Patterns *** '
'Training Pattern 1 -1 1 -1 -1 1 1 1 '
'Target Pattern 1 .3 .3 .3 .3 .3 .3 1 '
'Interpretation delay 0 '
' '
'Training Pattern 1 1 -1 1 -1 -1 1 1 '
'Target Pattern 1 .3 .3 .3 .3 .3 1 .3 '
'Interpretation delay 1 '
' '
'Training Pattern 1 1 1 -1 1 -1 -1 1 '
'Target Pattern 1 .3 .3 .3 .3 1 .3 .3 '
'Interpretation delay 2 '
' '
'Training Pattern 1 1 1 1 -1 1 -1 -1 '
'Target Pattern 1 .3 .3 .3 1 .3 .3 .3 '
'Interpretation delay 3 '
' '
'Training Pattern 1 -1 1 1 1 -1 1 -1 '
'Target Pattern 1 .3 .3 1 .3 .3 .3 .3 '
'Interpretation delay 4 '
' '
'Training Pattern 1 -1 -1 1 1 1 -1 1 '
'Target Pattern 1 .3 1 .3 .3 .3 .3 .3 '
'Interpretation delay 5 '
' '
'Training Pattern 1 1 -1 -1 1 1 1 -1 '
'Target Pattern 1 1 .3 .3 .3 .3 .3 .3 '
'Interpretation delay 6 '
' '
' This problem will be the demo problem for the Random Neural '
'Network Simulator program. For more information see the '
'following items in the help window: '
'Fast Start '
'Detailed Description '
'Parameters ---> Network File '
'Parameters ---> Train File '
'Parameters ---> Test File '};
ind = ind+1;
matList(ind).Name='Fast Start';
matList(ind).Help= {...
' Fast Start '
' ================== '
'1- Enter the name of the example network file by: '
' writing prob1_net.m in the Net File Name edit field or by '
' choosing the file prob1_net.m from the popup menu. '
'2- check off the Auto Save Weights (this is the check box '
' under the Weights File Name edit box(this step is important)'
'3- push the Accept button. '
'4- Push the Draw Net button to draw the network structure. '
'5- Push Train button to begin training. '
'6- Push Stop button to stop training after some iterations. '
'7- Push Plot Error button to plot the Mean Squared Error. '
'8- Push Resume button to continue training. '
'9- Wait until the network is trained (you will be indicated '
' via the output window. '
'10- Push Reset button to reset parameters and load another '
' file. '
'Note: * Of Course you can push the Help button in any time you '
' find this button enabled '
' * You should read the Detailed Description in the Help '
' window to learn more about the program usage. '};
ind = ind + 1;
matList(ind).Name = 'Detailed Description';
matList(ind).Help= {...
' Detailed Program Description '
' ============================= '
' When you launch the program, you will see two different '
'windows, the Random Neural Network (RNN) Simulator window and '
'the output window in the up right corner of the main window. '
'The RNN simulator window here is the main window which '
'will contain all the information needed to apply the RNN '
'training algorithm to a certain problem. The output window '
'acts as a log window which will contain messages to the user '
'indicating the validity of the current choice. You will notice '
'many buttons and edit boxes but only some are activated at a '
'certain time. This is made so as to minimize the confusion and '
'error possibility for the user since some buttons are likely '
'to cause an error when invoked with the improper sequence. You '
'will find the following user interface controls activated : '
'( Help) Push Button invoke the Help window '
'( About) Push Button invoke the About window '
'(RNN Tutorial) Push Button invoke the RNN tutorial window '
'(Quit) Push Button exit the RNN simulator '
'(Edit) Push Button edit the contents of the network '
' file '
'(Net File) Edit Box enter the network file name '
'(Net File) Popup Menu choose the network file name '
' Well, Now we want to go through some thing practical. To do '
'so you will find a set of files that will represent the problem '
'of code acquisition described in the Introduction choice in the '
'Help window. The files are : '
'Network File: (prob1_net.m) '
'Training File : (prob1_trn.m) '
'Testing File : (prob1_tst.m) '
'These three files are necessary for the proper operation of the '
'program. So your first step is to enter the network file name '
'(either by writing it in the Net File Name edit box and press '
'ENTER or by choosing it from the popup menu). '
' At this step you will see that the Train File Name Edit push '
'button and the Test File Name Edit push button are activated '
'now so that you can inspect the contents of these files. Note '
'that the programs tracks the typing errors so if you for example'
'try to enter a file name that does not exist, an error message '
'will appear on the output window (try to change the net file '
'name from prob1_net.m to rob1_net.m for example and see what '
'will happen) '
' If you check off the Auto Save Weights check box, you '
'will notice the following : '
' 1- you can change the number of hidden nodes, learning rate, '
'Output rate, Stop iteration, Stop MSE and Save iter edit boxes '
'and also initial weights range slide bar is enabled and you can '
'change the range of the initial random weights. '
'2- the loaded iteration is zero and the loaded MSE is None '
'3- The Reset button is enabled so that you can reset all the '
' parameters and begin a new session. '
'4- The Accept button is enabled so that you tell the program to '
' accept all the current parameters and begin training or '
' resume training '
'5- You can draw the current neural network structure by pressing'
' the Draw Net push button. '
' After pressing the Accept Push button, the program scans all'
'the current parameters and checks for errors, if it finds an '
'error error in any parameter, it will notify the user with what '
'parameter is in error. This notification appears in the output '
'window. If there is no errors, the program enters the training '
'phase( if the weights file is new this is the case here) or the '
'resuming/Testing phase (if the weight file is loaded from disk) '
' Now you can begin the training phase by pressing the Train '
'push button. The program now is in its training phase and you '
'will see the start iteration (0), Start MSE (None), Stop '
'iteration (300), Stop MSE (2e-3), current iteration, current '
'MSE, Iteration time and total elapsed time. '
'watch the current iteration until it reaches for example 33 '
'(and suppose that the current MSE at this point is .0047087 ) '
'and then press Stop push Button. '
' What we can do now? We can plot the error by pressing the '
'Plot Error push button, we can test the network at this point by'
'pressing the Test push button and then pressing the Show log '
'push button to show the network response to the testing patterns'
'or we can resume training by pressing the Resume push button '
'(and of course we can reset every thing by pressing reset) '
' If we press the Resume push button, we will return to the '
'training phase again but the starting iteration will be 33 and '
'the starting MSE is .0047087. During the training process you '
'will find a point at which the program will stop automatically '
'and a message will appear in the output window indicating '
'termination of the training process due to reaching the required'
'stop MSE value so you will see a message like: '
'Current Iteration = 129 '
'Stop Iteration = 300 '
'Current MSE = 0.001958206536 '
'Stop MSE = 0.002000000000 '
'Can not Resume because the Current MSE <= Stop MSE '
'A similar message will appear if the number of iterations is '
'exceeded. We call the current state : Over trained state '
' If you exit the program and then try to load an over trained'
'network, then you will not be able to train or resume training '
'So if you want to continue training, you should decrease the '
'Stop MSE parameter or increase the number of Stop Iterations. '
'Notice: When you first load a network file and make changes to '
'the parameters of the network, the changes will not be saved to '
'the same network file name but it will be saved to another file '
'with fixed name "temp_net.m" '
' '
' When you enter a network file name and you check on the '
'Auto Save Weights check box, the current parameters will be '
'loaded from the weights file instead of loading from the '
'network file. '
'Notice the auto saving bar that changes its color from Red '
'to Green every 15 iterations (the number of iterations after '
'which the program will automatically save the weights). '
' ------------------------------------------------------------- '};
ind = ind+1;
matList(ind).Name='Files';
matList(ind).Help= {...
' Files Usage '
' ==================== '};
ind=ind+1;
matList(ind).Name = ' Network File';
matList(ind).Help = {...
' Network File Name '
' ==================== '
' The name of the ( .m ) file that contains the network '
'parameters. A prototype for a network file is the file '
'prob1_net.m found in your directory The file must hav e '
'the flag NET_FLAG = 1 to indicate a valid network file '
'Also it must contain the other network parameters. '
'The contents of the prob1_net.m file is listed below: '
'( % infront of a line means commenting that line ) '
'---------------------------------------------------------'
'%%%% Network File Name: prob1_net.m %%% '
'NET_FLAG = 1; '
'%### Network Parameter Initialization ### '
'%Number of Input Neurons '
'N_Input = 7; '
'%Number of Hidden Neurons '
'N_Hidden = 8; '
'%Number of Output Neurons '
'N_Output = 7; '
'%Required Stop Mean Square Error value '
'Mse_Threshold = 0.000002; '
'%Learinig Rate '
'Eta = 0.1; '
'%Maximum Number of Iterations '
'N_Iterations = 300; '
'%Firing Rate of the Output Neurons '
'R_Out = 0.1; '
'%Random Range of the Weights '
'RAND_RANGE = .1; '
'%Flag for Fixing the Input Firing Rate '
'FIX_RIN = 1; '
' %Value of the Input Firing Rate(if Fixed) '
'R_IN = 0; '
'%No. Of Iterations for Auto Saving '
'N_Saved_Iterations = 15; '
'%Network File Name ( This File) '
'Net_File_Name = ''prob1_net.m''; '
'%Train Data FileName '
'Train_File_Name = ''prob1_trn.m''; '
'%Weights File Name '
'Weights_File_Name = ''prob1_wts.mat''; '
'%Test Data File Name '
'Test_File_Name = ''prob1_tst.m''; '
'%Output Log File Name '
'Log_File_Name = ''prob1_log.txt''; '
' '
' %%%%%%%%%%%%%%%%%%% '
'%Total Number of Neurons '
'N_Total = N_Input + N_Hidden + N_Output; '};
ind = ind+1;
matList(ind).Name = ' Train File';
matList(ind).Help = { ...
' Train File Name '
' ==================== '
' The name of the ( .m ) file that contains the training '
'and target patterns A prototype for a train file is the '
'file prob1_trn.m found in your directory The file must '
'have the flag TRN_FLAG = 1 to indicate a valid train data'
'file Also it must contain the number of training patterns'
'The contents of the prob1_net.m file is listed below: '
'( % infront of a line means commenting that line ) '
'---------------------------------------------------------'
'TRN_FLAG = 1; '
' %%% Training file: prob1.trn %%% '
'%Number of Patterns '
'N_Patterns = 7; '
'%Weights_File_Name = ''prob1_wts.mat''; '
'% ### Input Training Patterns ### '
'TRAIN_INPUT = [ '
' -1 1 -1 -1 1 1 1 '
' 1 -1 1 -1 -1 1 1 '
' 1 1 -1 1 -1 -1 1 '
' 1 1 1 -1 1 -1 -1 '
' -1 1 1 1 -1 1 -1 '
' -1 -1 1 1 1 -1 1 '
' 1 -1 -1 1 1 1 -1 '
']; '
'% ### Desired Output Patterns ### '
'TARGET = [ '
' .3 .3 .3 .3 .3 .3 1 '
' .3 .3 .3 .3 .3 1 .3 '
' .3 .3 .3 .3 1 .3 .3 '
' .3 .3 .3 1 .3 .3 .3 '
' .3 .3 1 .3 .3 .3 .3 '
' .3 1 .3 .3 .3 .3 .3 '
' 1 .3 .3 .3 .3 .3 .3 '
']; '
'%%%%%%%%%%%%% '};
ind = ind+1;
matList(ind).Name = ' Test File';
matList(ind).Help = { ...
' Test File Name '
' ==================== '
' The name of the ( .m ) file that contains the patterns'
'which we want to apply to the Random neural network and '
'record the response. A prototype for a test file is the '
'file prob1_tst.m found in your directory. The file must '
'have the flag TST_FLAG = 1 to indicate a valid test data '
'file. Also it must contain the number of testing patterns'
'The contents of the prob1_tst.m file is listed below: '
'(Notice that we train with 7 clear patters and we '
'test with 14 patters, the first 7 are the same as the '
'training patterns and the other 7 patterns are the first '
'7 patterns after adding to them gaussian noise such that '
'the signal to noise ratio is 2.2dB. '
'---------------------------------------------------------'
'TST_FLAG = 1; '
' %%% Testing file: prob1.tst %%% '
'N_Patterns = 14; %Number of Patterns '
'%Log_File_Name = ''prob1_log.txt''; '
'% ### Input Training Patterns ### '
'TEST_INPUT = [ '
' -1 1 -1 -1 1 1 1 '
' 1 -1 1 -1 -1 1 1 '
' 1 1 -1 1 -1 -1 1 '
' 1 1 1 -1 1 -1 -1 '
' -1 1 1 1 -1 1 -1 '
' -1 -1 1 1 1 -1 1 '
' 1 -1 -1 1 1 1 -1 '
' -1.313 1.157 -0.831 -1.149 0.791 1.638 1.766 '
' 1.062 -0.271 0.675 -0.762 -1.564 0.852 0.671 '
' 0.515 0.835 -1.800 0.841 -1.704 -1.910 1.156 '
' 1.408 0.920 1.643 -1.998 0.387 -0.994 -1.007 '
' -2.418 0.237 0.572 0.382 -1.241 1.042 -1.348 '
' -0.405 -1.998 0.993 1.145 1.104 -0.810 2.281 '
' 1.131 -1.422 -1.000 0.246 0.930 1.299 -1.154 '
']; '};
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
ind = ind+1;
matList(ind).Name = ' Weights File';
matList(ind).Help = { ...
' Weights File Name '
' ==================== '
'The name of the ( .mat) file that will contains the '
'network weights after being trained. A prototype for a '
'weights file is the file prob1_wts.m found in your '
'directory. '
'The weights may be saved automatically after a specific '
'number of iterations also the weights are saved after '
'pushing the Stop button and when the maximum number of '
'iterations is exceeded '
' When the Auto Save Weights check box is on, the '
'parameters will be initially loaded from the Weights file'
'and will be saved during training in the same weights '
'file. '
' When the Auto Save Weights check box is off, the '
'parameters will be initially loaded from the Network file'
'and will be saved during training to a temporary file '
'(temp_wts.mat). '};
ind = ind+1;
matList(ind).Name = ' Log File';
matList(ind).Help = { ...
' Log File Name '
' ==================== '
' The name of the ( .txt) file that will contains the '
'results of applying the testing patterns to the trained '
'Random neural network. A prototype for a log file is the'
'file prob1_log.txt that will be created in your directory'
'after testing the network. We applied the testing data '
'file that contains 14 patterns to the RNN after 21 '
'iterations and measure the response1 and after 83 '
'iterations and measure the respose2 : '
'( Notice the classification capability of the trained RNN'
'even when tested with very noisy versions of the inputs )'
' '
'Respose1 at Iteration No. 21 MSE = 0.074646000 '
'Respose1 at Iteration No. 83 MSE = 0.000001887 '
'-------------------------------------------- '
'pattern No. 1: '
'-1.0000 1.0000 -1.0000 -1.0000 1.0000 1.0000 1.0000'
' Response1: '
'0.2823 0.2995 0.3098 0.3119 0.3333 0.3006 0.9702'
' Response2: '
'0.3000 0.2999 0.3001 0.3004 0.3011 0.3001 0.9999'
' '
'pattern No. 2: '
'1.0000 -1.0000 1.0000 -1.0000 -1.0000 1.0000 1.0000'
' Response1: '
'0.3908 0.4096 0.3716 0.3114 0.4095 0.9729 0.3072'
' Response2: '
'0.3000 0.3000 0.3001 0.3005 0.3017 1.0000 0.2998'
' '
'pattern No. 3: '
'1.0000 1.0000 -1.0000 1.0000 -1.0000 -1.0000 1.0000'
' Response1: '
'0.4575 0.4770 0.1615 0.3326 0.9045 0.3072 0.3103'
' Response2: '
'0.2998 0.3000 0.2998 0.2999 1.0000 0.2998 0.2998'
' '
'pattern No. 4: '
'1.0000 1.0000 1.0000 -1.0000 1.0000 -1.0000 -1.0000'
' Response1: '
'0.4238 0.3322 0.3117 0.9903 0.2218 0.3021 0.2937'
' Response2: '
'0.2999 0.3003 0.3001 1.0000 0.2991 0.3000 0.2996'
' '
'pattern No. 5: '
'-1.0000 1.0000 1.0000 1.0000 -1.0000 1.0000 -1.0000'
' Response1: '
' 0.5386 0.5346 0.8066 0.2876 0.1114 0.3311 0.3298'
' Response2: '
' 0.3001 0.3001 1.0000 0.3003 0.3009 0.3000 0.3001'
' '
'pattern No. 6: '
'-1.0000 -1.0000 1.0000 1.0000 1.0000 -1.0000 1.0000'
' Response: '
' 0.2761 1.0000 0.3883 0.3173 0.3877 0.2779 0.2850'
' Response: '
' 0.3001 1.0000 0.3001 0.2997 0.2990 0.3000 0.3001'
' '
'pattern No. 7: '
'1.0000 -1.0000 -1.0000 1.0000 1.0000 1.0000 -1.0000'
' Response: '
'1.0000 0.2096 0.4071 0.3015 0.4018 0.2962 0.2670'
' Response: '
'1.0000 0.3000 0.3000 0.3002 0.3004 0.3001 0.3002'
' '
'pattern No. 8: '
'-1.3135 1.1570 -0.8315 -1.1497 0.7917 1.6385 1.7669'
' Response: '
' 0.2683 0.2996 0.3069 0.2805 0.3376 0.3206 0.9526'
' Response: '
' 0.2948 0.3033 0.2985 0.2952 0.3079 0.3061 1.0000'
' '
'pattern No. 9: '
'1.0621 -0.2719 0.6753 -0.7621 -1.5648 0.8528 0.6713'
' Response: '
'0.4396 0.3013 0.3154 0.2971 0.3957 0.8048 0.2605'
' Response: '
'0.3097 0.2524 0.2909 0.3230 0.2719 0.8918 0.2442'
' '
'pattern No. 10: '
'0.5154 0.8352 -1.8001 0.8416 -1.7047 -1.9105 1.1565'
' Response: '
'0.3249 0.5454 0.1573 0.2393 0.7778 0.2539 0.3522'
' Response: '
'0.2238 0.3109 0.3044 0.2519 0.8918 0.2619 0.3539'
' '
'pattern No. 11: '
'1.4083 0.9201 1.6436 -1.9984 0.3873 -0.9942 -1.0079'
' Response: '
'0.3795 0.3274 0.2990 0.8606 0.2223 0.3671 0.2076'
' Response: '
'0.2261 0.2259 0.3395 0.8951 0.3388 0.3382 0.2310'
' '
'pattern No. 12: '
'-2.4188 0.2377 0.5727 0.3826 -1.2410 1.0428 -1.3482'
' Response: '
' 0.4279 0.2886 0.6538 0.1406 0.0586 0.3875 0.2953'
' Response: '
' 0.2731 0.2156 0.6909 0.2126 0.1363 0.3922 0.2661'
' '
'pattern No. 13: '
'-0.4059 -1.9982 0.9932 1.1459 1.1042 -0.8101 2.2818'
' Response: '
' 0.2763 1.0000 0.3870 0.3163 0.3888 0.2769 0.2854'
' Response: '
' 0.2971 1.0000 0.2974 0.2971 0.3040 0.3054 0.3053'
' '
'pattern No. 14: '
'1.1313 -1.4221 -1.0005 0.2460 0.9303 1.2999 -1.1546'
' Response: '
'0.9491 0.1028 0.3195 0.3438 0.3159 0.3647 0.3191'
' Response: '
'0.8641 0.1946 0.2013 0.3698 0.1896 0.3667 0.3758'};
ind=ind+1;
matList(ind).Name='Parameters';
matList(ind).Help= { ...
' Network Parameters '};
ind=ind+1;
matList(ind).Name = ' Input Nodes';
matList(ind).Help= { ...
' Input Nodes '
' ==================== '
' Number of input nodes for the Random Neural Network. '
'This number of course depends on the type of the problem '
'under consideration and also it must be consistent with '
'the training and testing data patterns. The user can not '
' change the number of input nodes inside the window of '
'the program. '};
ind=ind+1;
matList(ind).Name = ' Hidden Nodes';
matList(ind).Help= { ...
' Hidden Nodes '
' ================== '
' Number of Hidden nodes for the Random Neural Network.'
'This number can be changed by the user to test the '
'effect of the number of hidden nodes on the performance. '};
ind=ind+1;
matList(ind).Name = ' Output Nodes';
matList(ind).Help= { ...
' Output Nodes '
' ================== '
' Number of output nodes for the Random Neural Network.'
'This number of course depend on the type of the problem '
'under consideration and also it must be consistent with '
'the target patterns. The user can not change the number '
'of output nodes inside the window of the program '};
ind=ind+1;
matList(ind).Name = ' Learning Rate';
matList(ind).Help= { ...
' Learning Rate '
' ================== '
' A real number between 0 and 1. '};
ind=ind+1;
matList(ind).Name = ' Output Rate';
matList(ind).Help= { ...
' Output Rate '
' ================== '
' The rate of the output neurons (refer the RNN '
'algorithm ) '};
ind=ind+1;
matList(ind).Name = ' Stop Iteration';
matList(ind).Help= { ...
' Stop Iteration '
' ================== '
' The number of iterations after which the training '
'will stop '};
ind=ind+1;
matList(ind).Name = ' Stop MSE';
matList(ind).Help= { ...
' Stop MSE '
' ================== '
' The value of the Mean Squared Error below which '
' training will stop '};
ind=ind+1;
matList(ind).Name = ' Save Iter';
matList(ind).Help= { ...
' Save Iter '
' ================== '
' The number of iterations after which the the program '
'will auto save the weights '};
ind=ind+1;
matList(ind).Name = ' Auto Save Weights';
matList(ind).Help= { ...
' Auto Save Weights '
' ================== '
'When it is checked on, Weights will be saved automatically'
'to the weights file. '};
ind=ind+1;
matList(ind).Name = ' Initial Weights Range';
matList(ind).Help= { ...
' Initial Weights Range '
' ================== '
' This is a number between 0 and 1. It defines the '
'upper range of uniform random numbers that will '
'be assigned initially to the weights '};
ind=ind+1;
matList(ind).Name = ' Loaded Iteration';
matList(ind).Help= { ...
' Loaded Iteration '
' ================== '
' The iteration number loaded from the weights file of '
'a pre trained network. If the weights file is created '
'then this value will be 0. '};
ind=ind+1;
matList(ind).Name = ' Loaded MSE';
matList(ind).Help= { ...
' Loaded MSE '
' ================== '
' The Mean Squares Error value loaded from the weights '
'file of a pre trained network. If the weights file is '
'created then this value will be None. '};
ind=ind+1;
matList(ind).Name = ' Start Iteration';
matList(ind).Help= { ...
' Start Iteration '
' ================== '
' The iteration number at which training will begin or '
'resumed '};
ind=ind+1;
matList(ind).Name = ' Current Iteration';
matList(ind).Help= { ...
' Current Iteration '
' ================== '
' The iteration number at the current training cycle. '};
ind=ind+1;
matList(ind).Name = ' Start MSE';
matList(ind).Help= { ...
' Start MSE '
' ================== '
' The Mean Squares Error value at which training '
'begins or resumes '};
ind=ind+1;
matList(ind).Name = ' Current MSE';
matList(ind).Help= { ...
' Current MSE '
' ================== '
' The Mean Squares Error value at the current training '
'cycle. '};
info=matList;
