find the value/s of the membership function when a crisp value is given???

Hi everyone;

I am not that much dealing with Matlab so my knowledge on it is quite limited. I want to work on Long Shor-Term Memory network (lstmlayer) function in matlab that uses for time series classification problems. As I can understand from the given definition of this function, the input is a cell array type and the output is a categorical type. My data is a crisp value (please see the attached mat file) data which contains a numerical data represented as 5 columns. Each column from column one to four representing a feature as; (start time, end time, duration and activity count for each class(column 5)), the last column has the class (output ) that I am going to classify using the LSTM. I did convert it to cell array data for the inputs and categorical data representing the classes in the outputs and used it as inputs and outputs to the lstm and it is working perfectly.

What I am looking for now is to fuzzify these crisp values (in the inputs) and representing them using any of the MFs like "gaussmf" and then use these fuzzified values as an input to the lstm network. Hopefully, by that, I gave you the idea that I want to apply. Please accept my apologies for wasting your time.

Thanks in advance

Gad

Answer 1

[Sam Chak]

Hi @Gadelhag M Omar Mohmed, It might be a little late answering this.

But a data-driven Mamdani Fuzzy System can be generated using the Fuzzy Clustering Method (FCM).

The Gaussian membership functions (gaussmf) are used for each FIS input.

load('set5_v73.mat');
inputData  = set5_v73(:, 1:4);
outputData = set5_v73(:, 5);

opt = genfisOptions('FCMClustering', 'FISType', 'mamdani');
fis = genfis(inputData, outputData, opt)

Iteration count = 1, obj. fcn = 157472.471121
Iteration count = 2, obj. fcn = 116043.324499
Iteration count = 3, obj. fcn = 103257.216462
Iteration count = 4, obj. fcn = 74286.012780
Iteration count = 5, obj. fcn = 42316.337039
Iteration count = 6, obj. fcn = 30912.483369
Iteration count = 7, obj. fcn = 22842.350019
Iteration count = 8, obj. fcn = 10819.232281
Iteration count = 9, obj. fcn = 8424.404455
Iteration count = 10, obj. fcn = 8256.121710
Iteration count = 11, obj. fcn = 8091.114189
Iteration count = 12, obj. fcn = 7985.259493
Iteration count = 13, obj. fcn = 7944.673590
Iteration count = 14, obj. fcn = 7922.418289
Iteration count = 15, obj. fcn = 7894.137043
Iteration count = 16, obj. fcn = 7846.677580
Iteration count = 17, obj. fcn = 7786.637470
Iteration count = 18, obj. fcn = 7747.018823
Iteration count = 19, obj. fcn = 7725.337083
Iteration count = 20, obj. fcn = 7707.933366
Iteration count = 21, obj. fcn = 7690.332424
Iteration count = 22, obj. fcn = 7671.331344
Iteration count = 23, obj. fcn = 7651.015348
Iteration count = 24, obj. fcn = 7630.114772
Iteration count = 25, obj. fcn = 7609.414216
Iteration count = 26, obj. fcn = 7589.150798
Iteration count = 27, obj. fcn = 7568.646365
Iteration count = 28, obj. fcn = 7546.245373
Iteration count = 29, obj. fcn = 7519.689536
Iteration count = 30, obj. fcn = 7487.783888
Iteration count = 31, obj. fcn = 7453.601524
Iteration count = 32, obj. fcn = 7424.010404
Iteration count = 33, obj. fcn = 7402.564985
Iteration count = 34, obj. fcn = 7387.715686
Iteration count = 35, obj. fcn = 7377.139635
Iteration count = 36, obj. fcn = 7369.449640
Iteration count = 37, obj. fcn = 7363.862772
Iteration count = 38, obj. fcn = 7359.849607
Iteration count = 39, obj. fcn = 7357.006426
Iteration count = 40, obj. fcn = 7355.018453
Iteration count = 41, obj. fcn = 7353.644538
Iteration count = 42, obj. fcn = 7352.704515
Iteration count = 43, obj. fcn = 7352.066860
Iteration count = 44, obj. fcn = 7351.637442
Iteration count = 45, obj. fcn = 7351.350003
Iteration count = 46, obj. fcn = 7351.158559
Iteration count = 47, obj. fcn = 7351.031566
Iteration count = 48, obj. fcn = 7350.947601
Iteration count = 49, obj. fcn = 7350.892226
Iteration count = 50, obj. fcn = 7350.855779
Iteration count = 51, obj. fcn = 7350.831825
Iteration count = 52, obj. fcn = 7350.816099
Iteration count = 53, obj. fcn = 7350.805783
Iteration count = 54, obj. fcn = 7350.799019
Iteration count = 55, obj. fcn = 7350.794585
Iteration count = 56, obj. fcn = 7350.791678
Iteration count = 57, obj. fcn = 7350.789772
Iteration count = 58, obj. fcn = 7350.788523
Iteration count = 59, obj. fcn = 7350.787703
Iteration count = 60, obj. fcn = 7350.787165
Iteration count = 61, obj. fcn = 7350.786812
Iteration count = 62, obj. fcn = 7350.786580
Iteration count = 63, obj. fcn = 7350.786427
Iteration count = 64, obj. fcn = 7350.786327
Iteration count = 65, obj. fcn = 7350.786261
Iteration count = 66, obj. fcn = 7350.786217
Iteration count = 67, obj. fcn = 7350.786188
Iteration count = 68, obj. fcn = 7350.786169
Iteration count = 69, obj. fcn = 7350.786157
Iteration count = 70, obj. fcn = 7350.786149

fis = 
  mamfis with properties:

                       Name: "mamdani41"
                  AndMethod: "min"
                   OrMethod: "max"
          ImplicationMethod: "min"
          AggregationMethod: "max"
      DefuzzificationMethod: "centroid"
    DisableStructuralChecks: 0
                     Inputs: [1×4 fisvar]
                    Outputs: [1×1 fisvar]
                      Rules: [1×10 fisrule]

	See 'getTunableSettings' method for parameter optimization.

% Plotting Fuzzy Inputs
figure(1)
subplot(2,2,1)
plotmf(fis, 'input', 1)
subplot(2,2,2)
plotmf(fis, 'input', 2)
subplot(2,2,3)
plotmf(fis, 'input', 3)
subplot(2,2,4)
plotmf(fis, 'input', 4)

% Plotting Fuzzy Output
figure(2)
plotmf(fis, 'output', 1)