source: cpp/gdk/neuroimpl-fuzzy.cpp @ 5

Last change on this file since 5 was 5, checked in by sz, 15 years ago

added the GDK (Genotype Development Kit)

File size: 7.9 KB
Line 
1// This file is a part of Framsticks GDK library.
2// Copyright (C) 2002-2006  Szymon Ulatowski.  See LICENSE.txt for details.
3// Refer to http://www.frams.alife.pl/ for further information.
4
5#include "neuroimpl-fuzzy.h"
6#include "neuroimpl-fuzzy-f0.h"
7
8int NI_FuzzyNeuro::countOuts(const Model *m, const Neuro *fuzzy)
9{
10  int outputs=0;
11  for(int i=0;i<m->getNeuroCount();i++)
12     for(int in=0;in<m->getNeuro(i)->getInputCount();in++)
13        if (m->getNeuro(i)->getInput(in)==fuzzy) outputs++;
14  return outputs;
15}
16
17int NI_FuzzyNeuro::lateinit()
18{
19  int i, maxOutputNr;
20
21  //check correctness of given parameters: string must not be null, sets&rules number > 0
22  if((fuzzySetsNr<1)||(rulesNr<1)||(fuzzySetString.len()==0)||(fuzzyRulesString.len()==0))
23    return 0; //error
24
25  // this part contains transformation of fuzzy sets
26  fuzzySets = new double[4*fuzzySetsNr]; //because every fuzzy set consist of 4 numbers
27  // converts fuzzy string from f0 to table of fuzzy numbers type 'double'
28  // (fill created space with numbers taken from string)
29  // also checks whether number of fuzzy sets in the string equals declared in the definition
30  if (FuzzyF0String::convertStrToSets(fuzzySetString, fuzzySets, fuzzySetsNr) != 0)
31    return 0; //error
32
33  // this part contains transformation of fuzzy rules and defuzzyfication parameters
34  rulesDef = new int[2*rulesNr];    //for each rule remembers number of inputs and outputs
35  //check correctness of string and fill in the rulesDef
36  if (FuzzyF0String::countInputsOutputs(fuzzyRulesString, rulesDef, rulesNr) == 0)
37  {
38    defuzzParam = new double[rulesNr]; // parameters used in defuzyfication process
39    // create space for rules according to rulesDef
40    rules = new int*[rulesNr];   //list of rules...
41    for (i=0; i<rulesNr; i++)    //...that contains rules body
42    {
43      rules[i] = new int[2*(rulesDef[2*i]+rulesDef[2*i+1])];  //each rule can have different number of inputs and outputs
44      defuzzParam[i] = 0; //should be done a little bit earlier, but why do not use this loop?
45    }
46    // fill created space with numbers taken from string
47    if (FuzzyF0String::convertStrToRules(fuzzyRulesString, rulesDef, rules, fuzzySetsNr, rulesNr, maxOutputNr) != 0)
48      return 0; //error
49  }
50  else
51    return 0; //error
52
53  setChannelCount(countOuts(neuro->owner, neuro));
54  return 1; //success
55}
56
57NI_FuzzyNeuro::~NI_FuzzyNeuro()
58{
59  if(rules) //delete rows and columns of **rules
60  {
61    for (int i=0; i<rulesNr; i++) SAFEDELETEARRAY(rules[i])
62    SAFEDELETEARRAY(rules)
63  }
64  SAFEDELETEARRAY(defuzzParam)
65  SAFEDELETEARRAY(rulesDef)
66  SAFEDELETEARRAY(fuzzySets)
67}
68
69int NI_FuzzyNeuro::GetFuzzySetParam(int set_nr, double &left, double &midleft, double &midright, double &right)
70{
71  if ( (set_nr>=0) && (set_nr<fuzzySetsNr) )
72  {
73    left = fuzzySets[4*set_nr];
74    midleft = fuzzySets[4*set_nr+1];
75    midright = fuzzySets[4*set_nr+2];
76    right = fuzzySets[4*set_nr+3];
77    return 0;
78  }
79  else
80    return 1;
81};
82
83/** Function conduct fuzzyfication of inputs and calculates - according to rules - crisp multi-channel output */
84void NI_FuzzyNeuro::go()
85{
86  if (Fuzzyfication()!=0)
87    return;
88  if (Defuzzyfication()!=0)
89    return;
90};
91
92/**
93* Function conduct fuzzyfication process - calculates minimum membership function (of every input) for each rule,
94* and writes results into defuzzParam - variable that contains data necessary for defuzzyfication
95*/
96int NI_FuzzyNeuro::Fuzzyfication()
97{
98  int i, j, nrIn, inputNr, nrFuzzySet;
99  double minimumCut; // actual minimal level of cut (= min. membership function)
100
101  // sets defuzzyfication parameters for each rule:
102  for (i=0; i<rulesNr; i++)
103  {
104    nrIn = rulesDef[2*i]; // nr of inputs in rule #i
105    minimumCut = 2; // the highest value of membership function is 1.0, so this value will definitely change
106    for (j=0; (j<nrIn)&&(minimumCut>0); j++) //minimumCut can not be <0, so if =0 then stop calculations
107    {
108      nrFuzzySet = rules[i][j*2 + 1]; // j*2 moves pointer through each output, +1 moves to nr of fuzzy set
109      inputNr = rules[i][j*2]; // as above but gives input number
110      minimumCut = min( minimumCut, TrapeziumFuzz(nrFuzzySet, getWeightedInputState(inputNr))); // value of membership function for this input and given fuzzy set
111    }
112    if ( (minimumCut>1) || (minimumCut<0) )
113      return 1;
114    defuzzParam[i] = minimumCut;
115  }
116  return 0;
117};
118
119/**
120* Function calculates value of the membership function of the set given by wchich_fuzzy_set for given crisp value input_val
121* In other words, this function fuzzyficates given crisp value with given fuzzy set, returning it's membership function
122* @param which_fuzzy_set - 0 < number of set < fuzzySetsNr
123* @param input_val - crisp value of input in range <-1; 1>
124* @return value of membership function (of given input for given set) in range <0;1> or, if error occur, negative value
125*/
126double NI_FuzzyNeuro::TrapeziumFuzz(int which_fuzzy_set, double input_val)
127{
128  double range=0, left=0, midleft=0, midright=0, right=0;
129
130  if ( (which_fuzzy_set < 0) || (which_fuzzy_set > fuzzySetsNr) )
131    return -2;
132  if ( (input_val < -1) || (input_val > 1) )
133    return -3;
134
135  if (GetFuzzySetParam(which_fuzzy_set, left, midleft, midright, right) != 0)
136    return -4;
137
138  if ( (input_val < left) || (input_val > right) ) // greather than right value
139    return 0;
140  else if ( (input_val >= midleft) && (input_val <= midright) ) // in the core of fuzzy set
141    return 1;
142  else if ( (input_val >= left) && (input_val < midleft) ) // at the left side of trapezium
143  {
144    range = fabs(midleft - left);
145    return fabs(input_val-left)/((range>0)?range:1); // quotient of distance between input and extreme left point of trapezium and range of rising side, or 1
146  }
147  else if ( (input_val > midright) && (input_val <= right) ) // at the right side of trapezium
148  {
149    range = fabs(right - midright);
150    return fabs(right-input_val)/((range>0)?range:1); // quotient of distance between input and extreme right point of trapezium and range of falling side, or 1
151  };
152
153  // should not occur
154  return 0;
155
156};
157
158/**
159* Function conducts defuzzyfication process: multi-channel output values are calculates with singleton method (method of high).
160* For each rules, all outputs fuzzy sets are taken and cut at 'cut-level', that is at minumum membership function (of current rule).
161* For all neuro pseudo-outputs, answer is calculated according to prior computations.
162* In fact, there is one output with multi-channel answer and appropriate values are given to right channels.
163*/
164int NI_FuzzyNeuro::Defuzzyfication()
165{
166  int i, j, nrIn, nrOut, out, set, outputsNr;
167  double *numerators, *denominators, midleft, midright, unimp;
168
169  outputsNr = getChannelCount();
170
171  numerators = new double[outputsNr];
172  denominators = new double[outputsNr];
173
174  for(i=0;i<outputsNr;i++) numerators[i] = denominators[i] = 0;
175
176  // for each rule...
177  for (i=0; i<rulesNr; i++)
178  {
179    nrIn = rulesDef[2*i]; // number of inputs in rule #i
180    nrOut = rulesDef[2*i + 1]; // number of outputs in rule #i
181    // ...calculate each output's product of middle fuzzy set value and minimum membership function (numerator) and sum of minimum membership function (denominator)
182    for (j=0; j<nrOut; j++)
183    {
184      out = rules[i][2*nrIn + 2*j]; //number of j-output
185      set = rules[i][2*nrIn + 2*j + 1]; //number of fuzzy set attributed to j-output
186      if (GetFuzzySetParam(set, unimp, midleft, midright, unimp) != 0) // gets range of core of given fuzzy set
187        { SAFEDELETEARRAY(denominators) SAFEDELETEARRAY(numerators) return 1; }
188      //defuzzParam[i] = minimum membership function for rule #i - calculated in fuzzyfication block
189      // defuzzyfication method of singletons (high): (fuzzy set modal value) * (minimum membership value)
190      numerators[out] += ((midleft + midright)/2.0) * defuzzParam[i];
191      denominators[out] += defuzzParam[i];
192    }
193  }
194
195  for (i=0; i<outputsNr; i++)
196  {
197    if (denominators[i] == 0)
198      setState(0, i);
199    else
200      setState(numerators[i]/denominators[i], i);
201  }
202
203  SAFEDELETEARRAY(denominators)
204  SAFEDELETEARRAY(numerators)
205
206  return 0;
207};
208
Note: See TracBrowser for help on using the repository browser.