Class: NeuralNetwork

Class Overview




Multi-layer Neural Network in PHP




Author(s):






    
                

Version:


    
  
  • 1.0
    • 



        






Methods
























Class Details



[line 75]

Multi-layer Neural Network in PHP

Loosely based on source code by Phil Brierley,  that was translated into PHP by 'dspink' in sep 2005
Algorithm was obtained from the excellent introductory book  "Artificial Intelligence - a guide to intelligent systems"  by Michael Negnevitsky (ISBN 0-201-71159-1)
Example: learning the 'XOR'-function  
  1.  require_once("class_neuralnetwork.php");
    2. 

  2.  
    3. 

  3.  // Create a new neural network with 3 input neurons,
    4. 

  4.  // 4 hidden neurons, and 1 output neuron
    5. 

  5.   $n = new NeuralNetwork(3, 4, 1);
    6. 

  6.  $n->setVerbose(false);
    7. 

  7.  
    8. 

  8.  // Add test-data to the network. In this case,
    9. 

  9.  // we want the network to learn the 'XOR'-function.
    10. 

  10.  // The third input-parameter is the 'bias'.
    11. 

  11.   $n->addTestData( array (-1, -1, 1), array (-1));
    12. 

  12.  $n->addTestData( array (-1,  1, 1), array ( 1));
    13. 

  13.  $n->addTestData( array ( 1, -1, 1), array ( 1));
    14. 

  14.  $n->addTestData( array ( 1,  1, 1), array (-1));
    15. 

  15.  
    16. 

  16.  // we try training the network for at most $max times
    17. 

  17.   $max = 3;
    18. 

  18.  
    19. 

  19.  // train the network in max 1000 epochs, with a max squared error of 0.01
    20. 

  20.   while (!($success=$n->train(1000, 0.01)) && $max-->0) {
    21. 

  21.          // training failed:
    22. 

  22.          // 1. re-initialize the weights in the network
    23. 

  23.                   $n->initWeights();
    24. 

  24.  
    25. 

  25.          // 2. display message
    26. 

  26.                   echo "Nothing found...<hr />";
    27. 

  27.  }
    28. 

  28.  
    29. 

  29.  // print a message if the network was succesfully trained
    30. 

  30.   if ($success) {
    31. 

  31.          $epochs = $n->getEpoch();
    32. 

  32.          echo "Success in $epochs training rounds!<hr />";
    33. 

  33.  }
    34. 

  34.  
    35. 

  35.  // in any case, we print the output of the neural network
    36. 

  36.   for ($i = 0; $i < count($n->trainInputs); $i ++) {
    37. 

  37.          $output = $n->calculate($n->trainInputs[$i]);
    38. 

  38.          print "<br />Testset $i; ";
    39. 

  39.          print "expected output = (".implode(", ", $n->trainOutput[$i]).") ";
    40. 

  40.          print "output from neural network = (".implode(", ", $output).")\n";
    41. 

  41.  }
    42. 

The resulting output could for example be something along the following lines:
  1.  Success in 719 training rounds!
    2. 

  2.  Testset 0; expected output = (-1) output from neural network = (-0.986415991978)
    3. 

  3.  Testset 1; expected output = (1) output from neural network = (0.992121412998)
    4. 

  4.  Testset 2; expected output = (1) output from neural network = (0.992469534962)
    5. 

  5.  Testset 3; expected output = (-1) output from neural network = (-0.990224120384)
    6. 

...which indicates the network has learned the task.





Tags:





  

    version:  1.0
  

  

    since:  feb 2007
  

  

    author:  E. Akerboom
  

  

    author:  Tremani, Delft
  

  

    license:  BSD License
  










[ Top ]








Class Methods




  

	
	
constructor NeuralNetwork [line 124]

	

    

    

		NeuralNetwork NeuralNetwork(
array
$nodecount)
    

    


	
		Creates a neural network.

Example:  
  1.  // create a network with 4 input nodes, 10 hidden nodes, and 4 output nodes
    2. 

  2.   $n = new NeuralNetwork(4, 10, 4);
    3. 

  3.  
    4. 

  4.  // create a network with 4 input nodes, 1 hidden layer with 10 nodes,
    5. 

  5.  // another hidden layer with 10 nodes, and 4 output nodes
    6. 

  6.   $n = new NeuralNetwork(4, 10, 10, 4);
    7. 

  7.  
    8. 

  8.  // alternative syntax
    9. 

  9.   $n = new NeuralNetwork(array(4, 10, 10, 4));
    10. 






	
        
Parameters:

    

    

          

        array  
        $nodecount  
        The number of nodes in the consecutive layers.
      

        

    


        
[ Top ]

  

  

	
	
method activation [line 234]

	

    

    

		float activation(
float
$value)
    

    


	
		Implements the standard (default) activation function for backpropagation networks,  the 'tanh' activation function.





Tags:





  

    return:  The final output of the node
  









	
        
Parameters:

    

    

          

        float  
        $value  
        The preliminary output to apply this function to
      

        

    


        
[ Top ]

  

  

	
	
method addControlData [line 294]

	

    

    

		void addControlData(
array
$input, array
$output, [int
$id = null])
    

    


	
		Add a set of control data to the network.

This set of data is used to prevent 'overlearning' of the network. The  network will stop training if the results obtained for the control data  are worsening.
The data added as control data is not used for training.





	
        
Parameters:

    

    

          

        array  
        $input  
        An input vector
      

          

        array  
        $output  
        The corresponding output
      

          

        int  
        $id  
        (optional) An identifier for this piece of data
      

        

    


        
[ Top ]

  

  

	
	
method addTestData [line 259]

	

    

    

		void addTestData(
array
$input, array
$output, [int
$id = null])
    

    


	
		Add a test vector and its output





	
        
Parameters:

    

    

          

        array  
        $input  
        An input vector
      

          

        array  
        $output  
        The corresponding output
      

          

        int  
        $id  
        (optional) An identifier for this piece of data
      

        

    


        
[ Top ]

  

  

	
	
method calculate [line 187]

	

    

    

		mixed calculate(
array
$input)
    

    


	
		Calculate the output of the neural network for a given input vector





Tags:





  

    return:  The output of the network
  









	
        
Parameters:

    

    

          

        array  
        $input  
        The vector to calculate
      

        

    


        
[ Top ]

  

  

	
	
method derivative_activation [line 246]

	

    

    

		$float derivative_activation(
float
$value)
    

    


	
		Implements the derivative of the activation function. By default, this is the  inverse of the 'tanh' activation function: 1.0 - tanh($value)*tanh($value);





	
        
Parameters:

    

    

          

        float  
        $value  
        'X'
      

        

    


        
[ Top ]

  

  

	
	
method getControlDataIDs [line 313]

	

    

    

		array getControlDataIDs(
)
    

    


	
		Returns the identifiers of the control data used during the training  of the network (if available)





Tags:





  

    return:  An array of identifiers
  









	
        
[ Top ]

  

  

	
	
method getLearningRate [line 155]

	

    

    

		float getLearningRate(
int
$layer)
    

    


	
		Gets the learning rate for a specific layer





Tags:





  

    return:  The learning rate for that layer
  









	
        
Parameters:

    

    

          

        int  
        $layer  
        The layer to obtain the learning rate for
      

        

    


        
[ Top ]

  

  

	
	
method getMomentum [line 177]

	

    

    

		float getMomentum(
)
    

    


	
		Gets the momentum.





Tags:





  

    return:  The momentum
  









	
        
[ Top ]

  

  

	
	
method getRandomWeight [line 647]

	

    

    

		float getRandomWeight(

$layer)
    

    


	
		Gets a random weight between [-0.25 .. 0.25]. Used to initialize the network.





Tags:





  

    return:  A random weight
  









	
        
Parameters:

    

    

          

          
        $layer  
        
      

        

    


        
[ Top ]

  

  

	
	
method getTestDataIDs [line 277]

	

    

    

		array getTestDataIDs(
)
    

    


	
		Returns the identifiers of the data used to train the network (if available)





Tags:





  

    return:  An array of identifiers
  









	
        
[ Top ]

  

  

	
	
method isVerbose [line 344]

	

    

    

		boolean isVerbose(
)
    

    


	
		Returns whether or not the network displays status and error messages.





Tags:





  

    return:  'true' if status and error messages are displayed, 'false' otherwise
  









	
        
[ Top ]

  

  

	
	
method load [line 355]

	

    

    

		boolean load(
string
$filename)
    

    


	
		Loads a neural network from a file saved by the 'save()' function. Clears  the training and control data added so far.





Tags:





  

    return:  'true' on success, 'false' otherwise
  









	
        
Parameters:

    

    

          

        string  
        $filename  
        The filename to load the network from
      

        

    


        
[ Top ]

  

  

	
	
method save [line 399]

	

    

    

		boolean save(
string
$filename)
    

    


	
		Saves a neural network to a file





Tags:





  

    return:  'true' on success, 'false' otherwise
  









	
        
Parameters:

    

    

          

        string  
        $filename  
        The filename to save the neural network to
      

        

    


        
[ Top ]

  

  

	
	
method setLearningRate [line 141]

	

    

    

		void setLearningRate(
array
$learningrate)
    

    


	
		Sets the learning rate between the different layers.





	
        
Parameters:

    

    

          

        array  
        $learningrate  
        An array containing the learning rates [range 0.0 - 1.0].  The size of this array is 'layercount - 1'. You might also provide a single number. If that is  the case, then this will be the learning rate for the whole network.
      

        

    


        
[ Top ]

  

  

	
	
method setMomentum [line 168]

	

    

    

		void setMomentum(
float
$momentum)
    

    


	
		Sets the 'momentum' for the learning algorithm. The momentum should  accelerate the learning process and help avoid local minima.





	
        
Parameters:

    

    

          

        float  
        $momentum  
        The momentum. Must be between 0.0 and 1.0; Usually between 0.5 and 0.9
      

        

    


        
[ Top ]

  

  

	
	
method setVerbose [line 335]

	

    

    

		void setVerbose(
boolean
$is_verbose)
    

    


	
		Determines if the neural network displays status and error messages. By default, it does.





	
        
Parameters:

    

    

          

        boolean  
        $is_verbose  
        'true' if you want to display status and error messages, 'false' if you don't
      

        

    


        
[ Top ]

  

  

	
	
method showWeights [line 322]

	

    

    

		void showWeights(
[boolean
$force = false])
    

    


	
		Shows the current weights and thresholds





	
        
Parameters:

    

    

          

        boolean  
        $force  
        Force the output, even if the network is not verbose.
      

        

    


        
[ Top ]

  

  

	
	
method train [line 424]

	

    

    

		bool train(
[int
$maxEpochs = 500], [float
$maxError = 0.01])
    

    


	
		Start the training process





Tags:





  

    return:  'true' if the training was successful, 'false' otherwise
  









	
        
Parameters:

    

    

          

        int  
        $maxEpochs  
        The maximum number of epochs
      

          

        float  
        $maxError  
        The maximum squared error in the training data
      

        

    


        
[ Top ]

  






        

		    

		    Documentation generated by phpDocumentor 1.3.1