Rank-1 NES algorithm

Signed-off-by: unknown <[email protected]>

examples/optimization/multiobjective/constnsga2jpq.py

@@ -0,0 +1,68 @@
+#!/usr/bin/env python
+""" An illustration of using the NSGA-II multi-objective optimization algorithm
+on Constrained Multi-Objective Optimization benchmark function. """
+
+__author__ = 'Jean Pierre Queau, [email protected]'
+
+from pybrain.optimization import ConstMultiObjectiveGA
+from pybrain.rl.environments.functions.multiobjective import ConstDeb,ConstSrn, \
+     ConstOsy,ConstTnk,ConstBnh
+import pylab
+from scipy import zeros, array
+
+# The Deb function
+#f = ConstDeb()
+# The Srinivas & Deb function
+#f = ConstSrn()
+# The Osyczka & Kundu function
+#f = ConstOsy()
+# The Tanaka function
+#f = ConstTnk()
+# The Binh & Korn function
+f = ConstBnh()
+# start at the origin
+x0 = zeros(f.indim)
+
+x0 = array([min_ for min_, max_ in f.xbound])
+
+# the optimization for a maximum of 25 generations
+n = ConstMultiObjectiveGA(f, x0, storeAllEvaluations = True, populationSize = 100, eliteProportion = 1.0,
+    topProportion = 1.0, mutationProb = 1.0, mutationStdDev = 0.3, storeAllPopulations = True, allowEquality = False)
+print 'Start Learning'
+n.learn(50)
+print 'End Learning'
+# plotting the results (blue = all evaluated points, red = resulting pareto front)
+print 'Plotting the Results'
+print 'All Evaluations.... take some time'
+for x in n._allEvaluations:
+    if x[1]:
+        pylab.plot([x[0][0]], [x[0][1]], 'b.')
+    else:
+        pylab.plot([x[0][0]], [x[0][1]], 'r.')
+for x in n.bestEvaluation: pylab.plot([x[0][0]], [x[0][1]], 'go')
+pylab.show()
+print 'Pareto Front'
+for x in n.bestEvaluation: pylab.plot([x[0][0]], [x[0][1]], 'go')
+pylab.show()
+
+print '==========='
+print '= Results =' 
+print '==========='
+'''
+i=0
+for gen in n._allGenerations:
+    print 'Generation: ',i
+    for j in range(len(gen[1])):
+        print gen[1].keys()[j],gen[1].values()[j]
+    i+=1
+'''
+print 'Population size ',n.populationSize
+print 'Elitism Proportion ',n.eliteProportion
+print 'Mutation Probability ',n.mutationProb
+print 'Mutation Std Deviation ',n.mutationStdDev
+print 'Objective Evaluation number ',n.numEvaluations
+print 'last generation Length of bestEvaluation ',len(n.bestEvaluation)
+print 'Best Evaluable : Best Evaluation'
+for i in range(len(n.bestEvaluation)):
+    assert len(n.bestEvaluation) == len(n.bestEvaluable)
+    print n.bestEvaluable[i],':',n.bestEvaluation[i]

examples/optimization/multiobjective/nsga2jpq.py

@@ -0,0 +1,58 @@
+#!/usr/bin/env python
+""" An illustration of using the NSGA-II multi-objective optimization algorithm
+on Unconstrained Multi-Objective Optimization benchmark function. """
+
+__author__ = 'Jean Pierre Queau, [email protected]'
+
+from pybrain.optimization import MultiObjectiveGA
+from pybrain.rl.environments.functions.multiobjective import Deb, Pol
+import pylab
+from scipy import zeros, array
+
+# The Deb function
+#f = Deb()
+# The Pol function
+f = Pol()
+
+# start at the origin
+x0 = zeros(f.indim)
+
+x0 = array([min_ for min_, max_ in f.xbound])
+
+# the optimization for a maximum of 25 generations
+n = MultiObjectiveGA(f, x0, storeAllEvaluations = True, populationSize = 50, eliteProportion = 1.0,
+    topProportion = 1.0, mutationProb = 0.5, mutationStdDev = 0.1, storeAllPopulations = True, allowEquality = False)
+print 'Start Learning'
+n.learn(30)
+print 'End Learning'
+
+# plotting the results (blue = all evaluated points, red = resulting pareto front)
+print 'Plotting the Results'
+print 'All Evaluations'
+for x in n._allEvaluations: pylab.plot([x[0]], [x[1]], 'b.')
+for x in n.bestEvaluation: pylab.plot([x[0]], [x[1]], 'ro')
+pylab.show()
+print 'Pareto Front'
+for x in n.bestEvaluation: pylab.plot([x[0]], [x[1]], 'ro')
+pylab.show()
+print '==========='
+print '= Results =' 
+print '==========='
+'''
+i=0
+for gen in n._allGenerations:
+    print 'Generation: ',i
+    for j in range(len(gen[1])):
+        print gen[1].keys()[j],gen[1].values()[j]
+    i+=1
+'''
+print 'Population size ',n.populationSize
+print 'Elitism Proportion ',n.eliteProportion
+print 'Mutation Probability ',n.mutationProb
+print 'Mutation Std Deviation ',n.mutationStdDev
+print 'Objective Evaluation number ',n.numEvaluations
+print 'last generation Length of bestEvaluation ',len(n.bestEvaluation)
+print 'Best Evaluable : Best Evaluation'
+for i in range(len(n.bestEvaluation)):
+    assert len(n.bestEvaluation) == len(n.bestEvaluable)
+    print n.bestEvaluable[i],':',n.bestEvaluation[i]

examples/optimization/multiobjective/readme.txt

@@ -0,0 +1,47 @@
+
+nsga2.py is the original example for the Kurbenchmark function
+
+nsga2jpq.py is the modified example to demonstrate the implementation
+            for Unconstrained Multi-Objective Optimization Problems and 
+            has been validated against the DEB by Deb (2001) function 
+            and the POL by Poloni & all (2000) function.
+            Modification have been made to the original code to implement
+            the boundaries of the parameters
+
+constnsga2jpq is new and is an example to demonstrate the implementation
+              for Constrained Multi-Objective Optimization Problems and 
+              has been validated against the DEB CONSTR by Deb (2001)
+              function, the SRN by Srinivas & Deb (1994) and the 
+              OSY by Osyczka and Kundu (1995).
+              New class has been defined and modification have been made
+              to the original code to implement the constrained functions.
+
+
+the modifications or new fuctions added are encapsulated in the code by
+
+""" added by JPQ """"
+
+  ....
+  ....
+
+# ---
+
+
+The following code files have been modified:
+
+pybrain/rl/environments/functions/multiobjective.py
+pybrain/rl/environments/functions/transformations.py
+pybrain/tools/nondominated.py
+pybrain/optimization/optimizer.py
+pybrain/optimization/populationbased/ga.py
+pybrain/optimization/populationbased/multiobjective/__init__.py
+pybrain/optimization/populationbased/multiobjective/nsga2.py
+pybrain/optimization/populationbased/multiobjective/constnsga2.py
+
+It is clear at the end that nsga2.py and constnsga2.py should be merged.
+In the transformation file only the oppositeFunction has been modified, may 
+be the other functions should also be modified but this was not required to 
+make the code running.
+
+Hope this will help to make you understanding what i have been doing.
+

examples/supervised/Test Network Reader&Writer/jpq2layersReader.py

@@ -0,0 +1,70 @@
+from pybrain.structure import FeedForwardNetwork
+from pybrain.tools.validation    import ModuleValidator,Validator
+from pybrain.utilities           import percentError
+from pybrain.tools.customxml     import NetworkReader
+from pybrain.datasets            import SupervisedDataSet
+import numpy
+import pylab
+import os
+
+def myplot(trns,ctrns = None,tsts = None,ctsts = None,iter = 0):
+  plotdir = os.path.join(os.getcwd(),'plot')
+  pylab.clf()
+  try:
+    assert len(tsts) > 1
+    tstsplot = True
+  except:
+    tstsplot = False
+  try:
+    assert len(ctsts) > 1
+    ctstsplot = True
+  except:
+    ctstsplot = False
+  try:
+    assert len(ctrns) > 1
+    ctrnsplot = True
+  except:
+    ctrnsplot = False
+  if tstsplot:
+    pylab.plot(tsts['input'],tsts['target'],c='b')
+  pylab.scatter(trns['input'],trns['target'],c='r')
+  if ctrnsplot:
+    pylab.scatter(trns['input'],ctrns,c='y')
+  if tstsplot and ctstsplot:
+    pylab.plot(tsts['input'], ctsts,c='g')
+
+  pylab.xlabel('x')
+  pylab.ylabel('y')
+  pylab.title('Neuron Number:'+str(nneuron))
+  pylab.grid(True)
+  plotname = os.path.join(plotdir,('jpq2layers_plot'+ str(iter)))
+  pylab.savefig(plotname)
+
+
+# set-up the neural network
+nneuron = 5
+mom = 0.98
+netname="LSL-"+str(nneuron)+"-"+str(mom)
+mv=ModuleValidator()
+v = Validator()
+
+
+#create the test DataSet
+x = numpy.arange(0.0, 1.0+0.01, 0.01)
+s = 0.5+0.4*numpy.sin(2*numpy.pi*x)
+tsts = SupervisedDataSet(1,1)
+tsts.setField('input',x.reshape(len(x),1))
+tsts.setField('target',s.reshape(len(s),1))
+#read the train DataSet from file
+trndata = SupervisedDataSet.loadFromFile(os.path.join(os.getcwd(),'trndata'))
+
+myneuralnet = os.path.join(os.getcwd(),'myneuralnet.xml')
+if os.path.isfile(myneuralnet):
+  n = NetworkReader.readFrom(myneuralnet,name=netname)
+  #calculate the test DataSet based on the trained Neural Network
+  ctsts = mv.calculateModuleOutput(n,tsts)
+  tserr = v.MSE(ctsts,tsts['target'])
+  print 'MSE error on TSTS:',tserr
+  myplot(trndata,tsts = tsts,ctsts = ctsts)
+
+  pylab.show()

examples/supervised/Test Network Reader&Writer/jpq2layersWriter.py

@@ -0,0 +1,130 @@
+from pybrain.structure import FeedForwardNetwork
+from pybrain.structure           import LinearLayer, SigmoidLayer
+from pybrain.structure           import BiasUnit,TanhLayer
+from pybrain.structure           import FullConnection
+from pybrain.datasets            import SupervisedDataSet
+from pybrain.supervised.trainers import BackpropTrainer, RPropMinusTrainer
+from pybrain.tools.validation    import ModuleValidator,Validator
+from pybrain.utilities           import percentError
+from pybrain.tools.customxml     import NetworkWriter
+import numpy
+import pylab
+import os
+
+def myplot(trns,ctrns,tsts = None,ctsts = None,iter = 0):
+  plotdir = os.path.join(os.getcwd(),'plot')
+  pylab.clf()
+  try:
+    assert len(tsts) > 1
+    tstsplot = True
+  except:
+    tstsplot = False
+  try:
+    assert len(ctsts) > 1
+    ctstsplot = True
+  except:
+    ctstsplot = False
+  if tstsplot:
+    pylab.plot(tsts['input'],tsts['target'],c='b')
+  pylab.scatter(trns['input'],trns['target'],c='r')
+  pylab.scatter(trns['input'],ctrns,c='y')
+  if tstsplot and ctstsplot:
+    pylab.plot(tsts['input'], ctsts,c='g')
+
+  pylab.xlabel('x')
+  pylab.ylabel('y')
+  pylab.title('Neuron Number:'+str(nneuron))
+  pylab.grid(True)
+  plotname = os.path.join(plotdir,('jpq2layers_plot'+ str(iter)))
+  pylab.savefig(plotname)
+
+
+# set-up the neural network
+nneuron = 5
+mom = 0.98
+netname="LSL-"+str(nneuron)+"-"+str(mom)
+mv=ModuleValidator()
+v = Validator()
+n=FeedForwardNetwork(name=netname)
+inLayer = LinearLayer(1,name='in')
+hiddenLayer = SigmoidLayer(nneuron,name='hidden0')
+outLayer = LinearLayer(1,name='out')
+biasinUnit = BiasUnit(name="bhidden0")
+biasoutUnit = BiasUnit(name="bout")
+n.addInputModule(inLayer)
+n.addModule(hiddenLayer)
+n.addModule(biasinUnit)
+n.addModule(biasoutUnit)
+n.addOutputModule(outLayer)
+in_to_hidden = FullConnection(inLayer,hiddenLayer)
+bias_to_hidden = FullConnection(biasinUnit,hiddenLayer)
+bias_to_out = FullConnection(biasoutUnit,outLayer)
+hidden_to_out = FullConnection(hiddenLayer,outLayer)
+n.addConnection(in_to_hidden)
+n.addConnection(bias_to_hidden)
+n.addConnection(bias_to_out)
+n.addConnection(hidden_to_out)
+
+n.sortModules()
+n.reset()
+
+#read the initail weight values from myparam2.txt
+filetoopen = os.path.join(os.getcwd(),'myparam2.txt')
+if os.path.isfile(filetoopen):
+  myfile = open('myparam2.txt','r')
+  c=[]
+  for line in myfile:
+    c.append(float(line))
+  n._setParameters(c)
+else:
+  myfile = open('myparam2.txt','w')
+  for i in n.params:
+    myfile.write(str(i)+'\n')
+myfile.close()
+
+#activate the neural networks
+act = SupervisedDataSet(1,1)
+act.addSample((0.2,),(0.880422606518061,))
+n.activateOnDataset(act)
+#create the test DataSet
+x = numpy.arange(0.0, 1.0+0.01, 0.01)
+s = 0.5+0.4*numpy.sin(2*numpy.pi*x)
+tsts = SupervisedDataSet(1,1)
+tsts.setField('input',x.reshape(len(x),1))
+tsts.setField('target',s.reshape(len(s),1))
+
+#read the train DataSet from file
+trndata = SupervisedDataSet.loadFromFile(os.path.join(os.getcwd(),'trndata'))
+
+#create the trainer
+
+t = BackpropTrainer(n, learningrate = 0.01 ,
+                    momentum = mom)
+#train the neural network from the train DataSet
+
+cterrori=1.0
+print "trainer momentum:"+str(mom)
+for iter in range(25):
+  t.trainOnDataset(trndata, 1000)
+  ctrndata = mv.calculateModuleOutput(n,trndata)
+  cterr = v.MSE(ctrndata,trndata['target'])
+  relerr = abs(cterr-cterrori)
+  cterrori = cterr
+  print 'iteration:',iter+1,'MSE error:',cterr
+  myplot(trndata,ctrndata,iter=iter+1)
+  if cterr < 1.e-5 or relerr < 1.e-7:
+    break
+#write the network using xml file     
+myneuralnet = os.path.join(os.getcwd(),'myneuralnet.xml')
+if os.path.isfile(myneuralnet):
+    NetworkWriter.appendToFile(n,myneuralnet)
+else:
+    NetworkWriter.writeToFile(n,myneuralnet)
+
+#calculate the test DataSet based on the trained Neural Network
+ctsts = mv.calculateModuleOutput(n,tsts)
+tserr = v.MSE(ctsts,tsts['target'])
+print 'MSE error on TSTS:',tserr
+myplot(trndata,ctrndata,tsts,ctsts)
+
+pylab.show()

examples/supervised/Test Network Reader&Writer/myparam2.txt

@@ -0,0 +1,16 @@
+-0.876387117735
+0.0228437748357
+0.0936056466341
+-0.90331132597
+-0.117868958091
+0.755572921676
+-0.48378563418
+-0.635899582263
+1.43586286085
+-0.791688941299
+1.16894753711
+1.47155607167
+0.301809828737
+0.814171589059
+1.02308464693
+-0.746364019039