featureName += '-FENS_dsf%i_dswl%i' % (
settingsDict['FENS Downsampling'],
settingsDict['FENS Window Length'])
# Assign a value to numFeatures
if settingsDict['NN Type'] is not None:
numFeatures = int(settingsDict['dA Num Hidden Units'])
else:
numFeatures = int(settingsDict['Number of Features'])
NCDlist = calculateNCDs(processPool,
featureName,
numFeatures,
int(settingsDict['Feature DownSample Factor']),
int(settingsDict['CRP Time Delay']),
int(settingsDict['CRP Dimension']),
settingsDict['CRP Method'],
settingsDict['CRP Neighbourhood Size'],
None,
None,
int(settingsDict['NCD Sequence Length']),
featureFileDict=featureFileDict,
pieceIds=pieceIds)
# Convert NCD files into a dataframe
runTime = str(datetime.now()).replace(':', '-')
MAPresult = None
if NCDlist is None: # there were errors e.g. in CRP calculation after downsampling
MAPresult = 0 # need to use something that is not None for the optimiser to find its best result
else:
dfNCDs = convertNCDs(NCDlist, dataFrameFileName=None)
# Get the overall MAP of the run and add to the setting
# Calculate NCDs
for key in setting.keys():
print key, ':', setting[key]
featureName = '%s-%s-%iv-%i-FENS_dsf%i_dswl%i' % (
baseFeatureName, NNtype, setting['dA Num Visible Units'],
setting['dA Num Hidden Units'],
setting['FENS Downsampling'],
setting['FENS Window Length'])
NCDlist = calculateNCDs(processPool,
featureName,
numFeatures,
setting['DownSample Factor'],
setting['Time Delay'],
setting['Dimension'],
CRPmethod,
setting['Neighbourhood Size'],
numFolders,
numFilesPerFolder,
setting['Sequence Length'],
featureFileDict=FENSfeatureFileDict,
pieceIds=pieceIds)
# Convert NCD files into a dataframe
runTime = str(datetime.now()).replace(':', '-')
MAPresult = None
if NCDlist is None: # there were errors e.g. in CRP calculation after downsampling
MAPresult = 0 # need to use something that is not None for the optimiser to find its best result
else:
dfNCDs = convertNCDs(NCDlist, dataFrameFileName=runTime)
# Get the overall MAP of the run and add to the setting
learningRateBoostFactor,
setting['dA Corruption Level'],
setting['dA Num Visible Units'],
setting['dA Num Hidden Units'], batchSize, freqStd,
numFolders, numFilesPerFolder, timeStacking)
else:
weightMatrix = biases = featureOffset = featureScaling = None
# Calculate NCDs
for key in setting.keys():
print key, ':', setting[key]
NCDlist = calculateNCDs(processPool, featureName, numFeatures,
setting['DownSample Factor'],
setting['Time Delay'],
setting['Dimension'], CRPmethod,
setting['Neighbourhood Size'],
numFolders, numFilesPerFolder,
setting['Sequence Length'],
weightMatrix, biases, featureOffset,
featureScaling, timeStacking)
# Convert NCD files into a dataframe
runTime = str(datetime.now()).replace(':', '-')
MAPresult = None
if NCDlist is None: # there were errors e.g. in CRP calculation after downsampling
MAPresult = 0 # need to use something that is not None for the optimiser to find its best result
else:
dfNCDs = convertNCDs(NCDlist, dataFrameFileName=runTime)
# Get the overall MAP of the run and add to the setting
MAPresult = getDataFrameMAPresult(dfNCDs)
if MAPresult is not None and MAPresult != 0:
featureScaling = 1
# Calculate NCDs
for key in settingsDict.keys():
print key, ':', settingsDict[key]
featureName = '%s-%s-%iv-%i' % (settingsDict['Feature Name'],
settingsDict['NN Type'],
numFeatures,
settingsDict['NN # Hidden Units'])
NCDlist = calculateNCDs(processPool = processPool,
featureName = settingsDict['Feature Name'],
numFeatures = numFeatures,
downSampleFactor = settingsDict['Feature DownSample Factor'],
timeDelay = settingsDict['CRP Time Delay'],
dimension = settingsDict['CRP Dimension'],
method = settingsDict['CRP Method'],
neighbourhoodSize = settingsDict['CRP Neighbourhood Size'],
numFolders = numFolders,
numFilesPerFolder = numFilesPerFolder,
sequenceLength = settingsDict['NCD Sequence Length'],
featureFileDict = transformedFeatureFileDict,
pieceIds = pieceIds)
# Convert NCD files into a dataframe
runTime = str(datetime.now()).replace(':', '-')
MAPresult = None
if NCDlist is None: # there were errors e.g. in CRP calculation after downsampling
MAPresult = 0 # need to use something that is not None for the optimiser to find its best result
else:
dfNCDs = convertNCDs(NCDlist, dataFrameFileName = runTime)
# Get the overall MAP of the run and add to the setting