PMS RI 2010/2011 Prof. dr Zikrija Avdagić, dipl.ing.el. ANFIS Editor GUI ANFIS Editor GUI

NEURAL NETWORK OVERFITTING P = -1:0.1: 1 training P = -1:0.01:1 checking generalisation Because of having more neurons (40) than training pairs(21) it will overfit our function in the process of aproximation.

anfisedit

ANFIS Editor GUI Example 1: Checking Data Helps Model Validation Loading data

The training data set The horizontal axis is marked data set index. This index indicates the row from which that input data value was obtained (whether or not the input is a vector or a scalar).

4 To load the checking data set from the workspace: a In the Load data portion of the GUI, select Checking in the Type column. b Click Load Data to open the Load from workspace dialog box. c Type fuzex1chkData as the variable name and click OK. The checking data appears in the GUI plot as pluses superimposed on the training data. When checking data is presented to anfis as well as training data, the FIS model is selected to have prameters associated with the minimum checking data model error

Initializing and Generating Your FIS Automatic FIS Structure Generation To initialize your FIS using anfis: 1 Choose Grid partition, the default partitioning method. The two partition methods, grid partitioning and subtractive clustering,. 2 Click on the Generate FIS button. Clicking this button displays a menu from which you can choose the number of membership functions, MFs, and the type of input and output membership functions. There are only two choices for the output membership function: constant and linear. This limitation of output membership function choices is because anfis only operates on Sugeno-type systems. 3 Fill in the entries as shown in the following figure, and click OK.

Viewing Your FIS Structure After you generate the FIS, you can view the model structure by clicking the Structure button in the middle of the right side of the GUI. A new GUI appears, as follows.

Prije treniranja

ANFIS Training (FIRST RESULT) The two anfis parameter optimization method options available for FIS training are hybrid (the default, mixed least squares and backpropagation) and backpropa (backpropagation). Error Tolerance is used to create a training stopping criterion, which is related to the error size. The training will stop after the training data error remains within this tolerance. This is best left set to 0 if you are unsure how your training error may behave. To start the training: 1 Leave the optimization method at hybrid. 2 Set the number of training epochs to 40, under the Epochs listing on the GUI. 3 Select Train Now. The following window appears on your screen. The plot shows the checking error as ♦ ♦ on the top. The training error appears as * * on the bottom.

Number of nodes: 16 Number of linear parameters: 6 Number of nonlinear parameters: 9 Total number of parameters: 15 Number of training data pairs: 25 Number of checking data pairs: 26 Number of fuzzy rules: 3 Start training ANFIS Number of nodes: 16 Number of linear parameters: 6 Number of nonlinear parameters: 9 Total number of parameters: 15 Number of training data pairs: 25 Number of checking data pairs: 26 Number of fuzzy rules: 3 End training ANFIS: 40 epoch

ANFIS Training (SECOND RESULT) The two anfis parameter optimization method options available for FIS training are hybrid (the default, mixed least squares and backpropagation) and backpropa (backpropagation). Error Tolerance is used to create a training stopping criterion, which is related to the error size. The training will stop after the training data error remains within this tolerance. This is best left set to 0 if you are unsure how your training error may behave. To start the training: 1 Leave the optimization method at hybrid. 2 Set the number of training epochs to 40, under the Epochs listing on the GUI. 3 Select Train Now. The following window appears on your screen. The plot shows the checking error as ♦ ♦ on the top. The training error appears as * * on the bottom. The checking error decreases up to a certain point in the training, and then it increases. This increase represents the point of model overfitting. anfis chooses the model parameters associated with the minimum checking error (just prior to this jump point). This example shows why the checking data option of anfis is useful.

fuzex1trnData fuzex1chkData Poslije teniranja

Model validation using testing and checking data Model validation is the process by which the input vectors from input/output data sets on which the FIS was not trained, are presented to the trained FIS model, to see how well the FIS model predicts the corresponding data set output values. This is acomplished with the ANFIS editor GUI using the so-called testing data set. We can also use another type of data set for model validation and that data validation set is referred to as the checkong data set and this set is used to conrol the potential for model overfitting the data. When checking data is presented to ANFIS as well as training data, the FIS model is selected to have parameters associated with the minimum checking data model error.

Test of FIS if we have had only training set

To test your FIS against the checking data, select Checking data in the Test FIS portion of the ANFIS Editor GUI, and click Test Now. When you test the checking data against the FIS, it looks satisfactory. Validation of FIS with checking data if we have had training and checking sets during the training process Testing of FIS with : a) training set b) testing set (generalisation) c) checking set are made on FIS which memberships parameters are associated with the minimum checking error if this errors indicate overfitting, It was shown during the training process when we have had training data against checking data.

Loading More Data with ANFIS If you load data into ANFIS after clearing previously loaded data, you must make sure that the newly loaded data sets have the same number of inputs as the previously loaded ones did. Otherwise, you must start a new anfisedit session from the command line. Without Checking Data Option and Clearing Data If you do not want to use the checking data option of ANFIS, then do not load any checking data before you train the FIS. If you decide to retrain your FIS with no checking data, you can unload the checking data in one of two ways: Select the Checking option button in the Load data portion of the ANFIS Editor GUI, and then click Clear Data to unload the checking data. Close the ANFIS Editor GUI, and go to the MATLAB command line, and retype anfisedit. In this case you must reload the training data. After clearing the data, you must regenerate your FIS. After the FIS is generated, you can use your first training experience to decide on the number of training epochs you want for the second round of training.

Questions 1.What is overfitting by neural network? 2.What is ANFIS? 3.Which learnig method can use ANFIS? 4.Why we use validation based on checking data? 5.What is diference between training data, test data and checking data?

20