'------------------------------------------------------------------
' License Notice:
'------------------------------------------------------------------
' All Rights Reserved - Anoop Madhusudanan,
' Mail: amazedsaint@gmail.com
' Website: http://amazedsaint.blogspot.com
' See my articles about BrainNet at
' http://amazedsaint-articles.blogspot.com for details
'
' You can use this code (or part of it), for non
' commercial and academic uses, as long as
' - You are keeping this notice along with it
' - You are not making any profit out of this
'------------------------------------------------------------------
Imports System.Xml.Serialization
Imports System.IO
Imports System.Xml
Imports System.Text
Namespace NeuralXML
'''<summary> Exception thrown when there is a problem in interpreter </summary>
Public Class InterpreterException
Inherits NeuralFrameworkException
Sub New(ByVal Message As String, ByVal e As Exception)
MyBase.New(Message, e)
End Sub
End Class
'''<summary>Neural XML Interpreter </summary>
Public Class NXMLInterpreter
'''<summary>Interpret a neural xml file</summary>
Private Function Load(ByVal file As String) As NeuralXML.DataModel.NXML
Try
''' Create an instance of the XmlSerializer specifying type and namespace.
Dim serializer As New XmlSerializer(GetType(NeuralXML.DataModel.NXML))
''' A FileStream is needed to read the XML document.
Dim fs As New FileStream(file, FileMode.Open)
Dim reader As New XmlTextReader(fs)
''' Declare an object variable of the type to be deserialized.
Dim nxmlDoc As NeuralXML.DataModel.NXML
''' Use the Deserialize method to restore the object'''s state.
nxmlDoc = serializer.Deserialize(reader)
reader.Close()
Return nxmlDoc
Catch ex As Exception
Throw New InterpreterException("Unable to load file. " & ex.Message, ex)
End Try
End Function
'''<summary>Save the nxml file</summary>
Private Sub Save(ByVal file As String, ByVal model As NeuralXML.DataModel.NXML)
Try
''' Create an instance of the XmlSerializer specifying type and namespace.
Dim serializer As New XmlSerializer(GetType(NeuralXML.DataModel.NXML))
''' To write to a file, create a StreamWriter object.
Dim myWriter As StreamWriter = New StreamWriter(file)
serializer.Serialize(myWriter, model)
myWriter.Close()
Catch ex As Exception
Throw New InterpreterException("Unable to save file. " & ex.Message, ex)
End Try
End Sub
'''<summary>Start Interpretation of the document</summary>
Public Sub Interpret(ByVal file As String)
'''Load the nxml document
Dim nxmlDoc As NeuralXML.DataModel.NXML = Load(file)
Dim serializer As NetworkSerializer = New NetworkSerializer()
Dim patternHelper As PatternProcessingHelper = New PatternProcessingHelper()
Dim imageHelper As ImageProcessingHelper = New ImageProcessingHelper()
Dim network As DataModel.Network
Dim datablock As DataModel.DataBlock
Dim pdata As DataModel.PatternData
Dim idata As DataModel.ImageData
Try
For Each network In nxmlDoc.NetworkCollection
Dim mynn As NeuralNetwork = New NeuralNetwork()
Dim helper As NetworkHelper = New NetworkHelper(mynn)
'''Loading the network
Dim filePath As String
serializer.LoadNetwork(network.LoadPath, mynn)
'''Iterate each datablock
For Each datablock In network.DataBlockCollection
'''Load the inputs
Dim input As ArrayList
Dim output As ArrayList
'''Process all pattern data
'''Loop to train the network
For Each pdata In datablock.PatternDataCollection
input = New ArrayList()
output = New ArrayList()
'''Let us start the training
If datablock.Type = DataModel.BlockType.Train Then
'''Get the input
Select Case pdata.InputType
Case DataModel.DataType.Array
'''Split this string to arraylist
Dim arraystr() As String = Split(pdata.InputValue.Trim(), ",")
Dim onestr As String
For Each onestr In arraystr
input.Add(CType(onestr, Single))
Next
Case DataModel.DataType.Number
input = patternHelper.ArrayListFromNumber(CType(pdata.InputValue.Trim(), Single), mynn.InputLayer.Count)
Case DataModel.DataType.Pattern
input = patternHelper.ArrayListFromPattern(pdata.InputValue.Trim)
Case DataModel.DataType.Char
input = patternHelper.ArrayListFromChar(pdata.InputValue.Trim)
End Select
'''Get the output
Select Case pdata.OutputType
Case DataModel.DataType.Array
'''Split this string to arraylist
Dim arraystr() As String = Split(pdata.OutputValue.Trim(), ",")
Dim onestr As String
For Each onestr In arraystr
output.Add(CType(onestr, Single))
Next
Case DataModel.DataType.Number
output = patternHelper.ArrayListFromNumber(CType(pdata.OutputValue.Trim, Long), mynn.OutputLayer.Count)
Case DataModel.DataType.Pattern
output = patternHelper.ArrayListFromPattern(pdata.OutputValue.Trim)
Case DataModel.DataType.Char
output = patternHelper.ArrayListFromChar(pdata.OutputValue.Trim)
End Select
PrintArr("Train: Input: ", input)
PrintArr("Train: Output: ", output)
helper.AddTrainingData(input, output)
End If
Next pdata
'''Loop all image data
'''Loop to train the network
For Each idata In datablock.ImageDataCollection
input = New ArrayList()
output = New ArrayList()
'''Let us start the training
If datablock.Type = DataModel.BlockType.Train Then
'''Get the input
Dim bmapInput As New Bitmap(idata.InputFile)
If Not (bmapInput.Width = idata.InputWidth And bmapInput.Height = idata.InputHeight) Then
bmapInput = imageHelper.ShrinkImage(bmapInput, idata.InputWidth, idata.InputHeight, True)
End If
input = imageHelper.ArrayListFromImage(bmapInput)
'''Get the output
If idata.OutputFile <> "" Then
Dim bmapOutput As New Bitmap(idata.OutputFile)
If Not (bmapOutput.Width = idata.OutputWidth And bmapOutput.Height = idata.OutputHeight) Then
bmapOutput = imageHelper.ShrinkImage(bmapOutput, idata.OutputWidth, idata.OutputHeight, True)
End If
output = imageHelper.ArrayListFromImage(bmapOutput)
Else
'''Get the output
Select Case idata.OutputType
Case DataModel.DataType.Array
'''Split this string to arraylist
Dim arraystr() As String = Split(idata.OutputValue.Trim(), ",")
Dim onestr As String
For Each onestr In arraystr
output.Add(CType(onestr, Single))
Next
Case DataModel.DataType.Number
output = patternHelper.ArrayListFromNumber(CType(idata.OutputValue.Trim(), Long), mynn.OutputLayer.Count)
Case DataModel.DataType.Pattern
output = patternHelper.ArrayListFromPattern(idata.OutputValue.Trim)
Case DataModel.DataType.Char
output = patternHelper.ArrayListFromChar(idata.OutputValue.Trim)
End Select
PrintArr("Train: Input: ", input)
PrintArr("Train: Output: ", output)
End If
helper.AddTrainingData(input, output)
End If
Next idata
'''Train the network
Console.WriteLine("Training Or Running Network...")
AddHandler helper.TrainingProgress, AddressOf Me.TrainingProgress
helper.Train(datablock.TrainCount)
'''Loop to run the network
For Each pdata In datablock.PatternDataCollection
input = New ArrayList()
output = New ArrayList()
If datablock.Type = DataModel.BlockType.Run Then
'''Get the input
Select Case pdata.InputType
Case DataModel.DataType.Array
'''Split this string to arraylist
Dim arraystr() As String = Split(pdata.InputValue.Trim(), ",")
Dim onestr As String
For Each onestr In arraystr
input.Add(CType(onestr, Single))
Next
Case DataModel.DataType.Number
input = patternHelper.ArrayListFromNumber(CType(pdata.InputValue.Trim(), Single), mynn.InputLayer.Count)
Case DataModel.DataType.Pattern
input = patternHelper.ArrayListFromPattern(pdata.InputValue.Trim)
Case DataModel.DataType.Char
input = patternHelper.ArrayListFromChar(pdata.InputValue.Trim)
End Select
mynn.RunNetwork(input)
PrintArr("Run: Input:", input)
'''Show the output
output = mynn.GetOutput()
PrintArr("Run: Output:", output)
Dim outstr As String = ""
Select Case pdata.OutputType
Case DataModel.DataType.Array
'''Split this string to arraylist
Dim acount As Long
For acount = 0 To output.Count - 1
outstr = outstr & CType(output(acount), String)
If acount < output.Count - 1 Then outstr = outstr & ","
Next
Case DataModel.DataType.Number
outstr = CType(patternHelper.NumberFromArraylist(output), String)
Case DataModel.DataType.Pattern
outstr = patternHelper.PatternFromArraylist(output)
Case DataModel.DataType.Char
outstr = patternHelper.CharFromArraylist(output)
End Select
pdata.OutputValue = outstr
End If
Next pdata
'''Loop to run the network
For Each idata In datablock.ImageDataCollection
input = New ArrayList()
output = New ArrayList()
If datablock.Type = DataModel.BlockType.Run Then
'''Get the input
Dim bmapInput As New Bitmap(idata.InputFile)
If Not (bmapInput.Width = idata.InputWidth And bmapInput.Height = idata.InputHeight) Then
bmapInput = imageHelper.ShrinkImage(bmapInput, idata.InputWidth, idata.InputHeight, True)
End If
input = imageHelper.ArrayListFromImage(bmapInput)
mynn.RunNetwork(input)
'''Show the output
output = mynn.GetOutput()
PrintArr("Run: Output:", output)
Dim outstr As String = ""
'''Save the output as an image file if required
If idata.OutputFile <> "" Then
Dim bmapOutput As Bitmap = imageHelper.ImageFromArraylist(output, idata.OutputWidth, idata.OutputHeight)
bmapoutput.Save(idata.OutputFile)
End If
Select Case idata.OutputType
Case DataModel.DataType.Array
'''Split this string to arraylist
Dim acount As Long
For acount = 0 To output.Count - 1
outstr = outstr & CType(output(acount), String)
If acount < output.Count - 1 Then outstr = outstr & ","
Next
Case DataModel.DataType.Number
outstr = CType(patternHelper.NumberFromArraylist(output), String)
Case DataModel.DataType.Pattern
outstr = patternHelper.PatternFromArraylist(output)
Case DataModel.DataType.Char
outstr = patternHelper.CharFromArraylist(output)
End Select
idata.OutputValue = outstr
End If
Next idata
Next datablock
'''Save the network
If network.SavePath.Trim = "" Then network.SavePath = network.LoadPath
serializer.SaveNetwork(network.SavePath, mynn)
Next network
Catch ex As Exception
Throw New InterpreterException("Unable to interpret. Invalid file path provided for image data? " & ex.Message, ex)
End Try
'''Update this nxml file
Save(file, nxmlDoc)
End Sub
Private Sub TrainingProgress(ByVal Current As Long, ByVal Max As Long, ByRef cancel As Boolean)
Console.Write(".")
End Sub
End Class
End Namespace
Submit an article or tip