ApicalTMPairMonitorMixin)
class MonitoredApicalTiebreakPairMemory(
ApicalTMPairMonitorMixin, ApicalTiebreakPairMemory):
pass
maxLength = 20
numTrainSequence = 50
numTestSequence = 50
rptPerSeq = 5
n = 2048
w = 40
enc = SDRCategoryEncoder(n, w, categoryList)
sdr_dict = dict()
for i in xrange(len(categoryList)):
sdr = enc.encode(categoryList[i])
activeCols = set(np.where(sdr)[0])
sdr_dict[categoryList[i]] = activeCols
def initializeTM():
DEFAULT_TEMPORAL_MEMORY_PARAMS = {"columnDimensions": (2048,),
"cellsPerColumn": 32,
"activationThreshold": 15,
"initialPermanence": 0.41,
"connectedPermanence": 0.5,
"minThreshold": 10,
"maxNewSynapseCount": 20,
def getDescription(datasets):
# ========================================================================
# Network definition
# Encoder for the sensor
encoder = MultiEncoder()
if 'filenameCategory' in datasets:
categories = [x.strip() for x in open(datasets['filenameCategory'])]
else:
categories = [chr(x + ord('a')) for x in range(26)]
if config['overlappingPatterns']:
encoder.addEncoder(
"name",
SDRCategoryEncoder(n=200,
w=config['spNumActivePerInhArea'],
categoryList=categories,
name="name"))
else:
encoder.addEncoder(
"name",
CategoryEncoder(w=config['spNumActivePerInhArea'],
categoryList=categories,
name="name"))
# ------------------------------------------------------------------
# Node params
# The inputs are long, horizontal vectors
inputDimensions = (1, encoder.getWidth())
# Layout the coincidences vertically stacked on top of each other, each
# looking at the entire input field.
columnDimensions = (config['spCoincCount'], 1)
# If we have disableSpatial, then set the number of "coincidences" to be the
# same as the encoder width
if config['disableSpatial']:
columnDimensions = (encoder.getWidth(), 1)
config['trainSP'] = 0
sensorParams = dict(
# encoder/datasource are not parameters so don't include here
verbosity=config['sensorVerbosity'])
CLAParams = dict(
# SP params
disableSpatial=config['disableSpatial'],
inputDimensions=inputDimensions,
columnDimensions=columnDimensions,
potentialRadius=inputDimensions[1] / 2,
potentialPct=1.00,
gaussianDist=0,
commonDistributions=0, # should be False if possibly not training
localAreaDensity=-1, #0.05,
numActiveColumnsPerInhArea=config['spNumActivePerInhArea'],
dutyCyclePeriod=1000,
stimulusThreshold=1,
synPermInactiveDec=0.11,
synPermActiveInc=0.11,
synPermActiveSharedDec=0.0,
synPermOrphanDec=0.0,
minPctDutyCycleBeforeInh=0.001,
minPctDutyCycleAfterInh=0.001,
spVerbosity=config['spVerbosity'],
spSeed=1,
printPeriodicStats=int(config['spPrintPeriodicStats']),
# TM params
tpSeed=1,
disableTemporal=0 if config['trainTP'] else 1,
temporalImp=config['temporalImp'],
nCellsPerCol=config['tpNCellsPerCol'] if config['trainTP'] else 1,
collectStats=1,
burnIn=2,
verbosity=config['tpVerbosity'],
newSynapseCount=config['spNumActivePerInhArea'],
minThreshold=config['spNumActivePerInhArea'],
activationThreshold=config['spNumActivePerInhArea'],
initialPerm=config['tpInitialPerm'],
connectedPerm=0.5,
permanenceInc=config['tpPermanenceInc'],
permanenceDec=config['tpPermanenceDec'], # perhaps tune this
globalDecay=config['tpGlobalDecay'],
pamLength=config['tpPAMLength'],
maxSeqLength=config['tpMaxSeqLength'],
maxAge=config['tpMaxAge'],
# General params
computeTopDown=config['computeTopDown'],
trainingStep='spatial',
)
dataSource = FileRecordStream(datasets['filenameTrain'])
description = dict(
options=dict(logOutputsDuringInference=False, ),
network=dict(sensorDataSource=dataSource,
sensorEncoder=encoder,
sensorParams=sensorParams,
CLAType='py.CLARegion',
CLAParams=CLAParams,
classifierType=None,
classifierParams=None),
)
if config['trainSP']:
description['spTrain'] = dict(
iterationCount=config['iterationCountTrain'],
#iter=displaySPCoincidences(50),
#finish=printSPCoincidences()
),
else:
description['spTrain'] = dict(
# need to train with one iteration just to initialize data structures
iterationCount=1)
if config['trainTP']:
description['tpTrain'] = []
for i in range(config['trainTPRepeats']):
stepDict = dict(
name='step_%d' % (i),
setup=sensorRewind,
iterationCount=config['iterationCountTrain'],
)
if config['tpTimingEvery'] > 0:
stepDict['iter'] = printTPTiming(config['tpTimingEvery'])
stepDict['finish'] = [printTPTiming(), printTPCells]
description['tpTrain'].append(stepDict)
# ----------------------------------------------------------------------------
# Inference tests
inferSteps = []
if config['evalTrainingSetNumIterations'] > 0:
# The training set. Used to train the n-grams.
inferSteps.append(
dict(
name='confidenceTrain_baseline',
iterationCount=min(config['evalTrainingSetNumIterations'],
config['iterationCountTrain']),
ppOptions=dict(
verbosity=config['ppVerbosity'],
printLearnedCoincidences=True,
nGrams='train',
#ipsDetailsFor = "name,None,2",
),
#finish=printTPCells,
))
# Testing the training set on both the TM and n-grams.
inferSteps.append(
dict(
name='confidenceTrain_nonoise',
iterationCount=min(config['evalTrainingSetNumIterations'],
config['iterationCountTrain']),
setup=[sensorOpen(datasets['filenameTrain'])],
ppOptions=dict(
verbosity=config['ppVerbosity'],
printLearnedCoincidences=False,
nGrams='test',
burnIns=[1, 2, 3, 4],
#ipsDetailsFor = "name,None,2",
#ipsAt = [1,2,3,4],
),
))
# The test set
if True:
if datasets['filenameTest'] != datasets['filenameTrain']:
inferSteps.append(
dict(
name='confidenceTest_baseline',
iterationCount=config['iterationCountTest'],
setup=[sensorOpen(datasets['filenameTest'])],
ppOptions=dict(
verbosity=config['ppVerbosity'],
printLearnedCoincidences=False,
nGrams='test',
burnIns=[1, 2, 3, 4],
#ipsAt = [1,2,3,4],
ipsDetailsFor="name,None,2",
),
))
description['infer'] = inferSteps
return description
def testMultiEncoder(self):
"""Testing MultiEncoder..."""
e = MultiEncoder()
# should be 7 bits wide
# use of forced=True is not recommended, but here for readibility, see
# scalar.py
e.addEncoder(
"dow",
ScalarEncoder(w=3,
resolution=1,
minval=1,
maxval=8,
periodic=True,
name="day of week",
forced=True))
# sould be 14 bits wide
e.addEncoder(
"myval",
ScalarEncoder(w=5,
resolution=1,
minval=1,
maxval=10,
periodic=False,
name="aux",
forced=True))
self.assertEqual(e.getWidth(), 21)
self.assertEqual(e.getDescription(), [("day of week", 0), ("aux", 7)])
d = DictObj(dow=3, myval=10)
expected = numpy.array([0, 1, 1, 1, 0, 0, 0] +
[0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1],
dtype="uint8")
output = e.encode(d)
self.assertTrue(numpy.array_equal(expected, output))
# Check decoding
decoded = e.decode(output)
self.assertEqual(len(decoded), 2)
(ranges, _) = decoded[0]["aux"]
self.assertEqual(len(ranges), 1)
self.assertTrue(numpy.array_equal(ranges[0], [10, 10]))
(ranges, _) = decoded[0]["day of week"]
self.assertTrue(
len(ranges) == 1 and numpy.array_equal(ranges[0], [3, 3]))
e.addEncoder(
"myCat",
SDRCategoryEncoder(n=7,
w=3,
categoryList=["run", "pass", "kick"],
forced=True))
d = DictObj(dow=4, myval=6, myCat="pass")
output = e.encode(d)
topDownOut = e.topDownCompute(output)
self.assertAlmostEqual(topDownOut[0].value, 4.5)
self.assertEqual(topDownOut[1].value, 6.0)
self.assertEqual(topDownOut[2].value, "pass")
self.assertEqual(topDownOut[2].scalar, 2)
self.assertEqual(topDownOut[2].encoding.sum(), 3)
def getDescription(datasets):
# ========================================================================
# Encoder for the sensor
encoder = MultiEncoder()
if config['encodingFieldStyleA'] == 'contiguous':
encoder.addEncoder('fieldA', ScalarEncoder(w=config['encodingOnBitsA'],
n=config['encodingFieldWidthA'], minval=0,
maxval=config['numAValues'], periodic=True, name='fieldA'))
elif config['encodingFieldStyleA'] == 'sdr':
encoder.addEncoder('fieldA', SDRCategoryEncoder(w=config['encodingOnBitsA'],
n=config['encodingFieldWidthA'],
categoryList=range(config['numAValues']), name='fieldA'))
else:
assert False
if config['encodingFieldStyleB'] == 'contiguous':
encoder.addEncoder('fieldB', ScalarEncoder(w=config['encodingOnBitsB'],
n=config['encodingFieldWidthB'], minval=0,
maxval=config['numBValues'], periodic=True, name='fieldB'))
elif config['encodingFieldStyleB'] == 'zero':
encoder.addEncoder('fieldB', SDRRandomEncoder(w=0, n=config['encodingFieldWidthB'],
name='fieldB'))
elif config['encodingFieldStyleB'] == 'sdr':
encoder.addEncoder('fieldB', SDRCategoryEncoder(w=config['encodingOnBitsB'],
n=config['encodingFieldWidthB'],
categoryList=range(config['numBValues']), name='fieldB'))
else:
assert False
# ========================================================================
# Network definition
# ------------------------------------------------------------------
# Node params
# The inputs are long, horizontal vectors
inputShape = (1, encoder.getWidth())
# Layout the coincidences vertically stacked on top of each other, each
# looking at the entire input field.
coincidencesShape = (config['spCoincCount'], 1)
inputBorder = inputShape[1]/2
if inputBorder*2 >= inputShape[1]:
inputBorder -= 1
sensorParams = dict(
# encoder/datasource are not parameters so don't include here
verbosity=config['sensorVerbosity']
)
CLAParams = dict(
inputShape = inputShape,
inputBorder = inputBorder,
coincidencesShape = coincidencesShape,
coincInputRadius = inputShape[1]/2,
coincInputPoolPct = 1.0,
gaussianDist = 0,
commonDistributions = 0, # should be False if possibly not training
localAreaDensity = -1, #0.05,
numActivePerInhArea = config['spNumActivePerInhArea'],
dutyCyclePeriod = 1000,
stimulusThreshold = 1,
synPermInactiveDec = config['spSynPermInactiveDec'],
synPermActiveInc = 0.02,
synPermActiveSharedDec=0.0,
synPermOrphanDec = 0.0,
minPctDutyCycleBeforeInh = 0.001,
minPctDutyCycleAfterInh = config['spMinPctDutyCycleAfterInh'],
minDistance = 0.05,
computeTopDown = 1,
spVerbosity = config['spVerbosity'],
spSeed = 1,
printPeriodicStats = int(config['spPeriodicStats']),
# TP params
disableTemporal = 1,
# General params
trainingStep = 'spatial',
)
trainingDataSource = FileRecordStream(datasets['trainingFilename'])
description = dict(
options = dict(
logOutputsDuringInference = False,
),
network = dict(
sensorDataSource = trainingDataSource,
sensorEncoder = encoder,
sensorParams = sensorParams,
CLAType = 'py.CLARegion',
CLAParams = CLAParams,
classifierType = None,
classifierParams = None),
)
if config['trainSP']:
description['spTrain'] = dict(
iterationCount=config['iterationCount'],
#iter=displaySPCoincidences(50),
finish=printSPCoincidences()
),
else:
description['spTrain'] = dict(
# need to train with one iteration just to initialize data structures
iterationCount=1)
# ============================================================================
# Inference tests
inferSteps = []
# ----------------------------------------
# Training dataset
if True:
datasetName = 'bothTraining'
inferSteps.append(
dict(name = '%s_baseline' % datasetName,
iterationCount = config['iterationCount'],
setup = [sensorOpen(datasets['trainingFilename'])],
ppOptions = dict(printLearnedCoincidences=True),
)
)
inferSteps.append(
dict(name = '%s_acc' % datasetName,
iterationCount = config['iterationCount'],
setup = [sensorOpen(datasets['trainingFilename'])],
ppOptions = dict(onlyClassificationAcc=True,
tpActivationThresholds=config['tpActivationThresholds'],
computeDistances=True,
verbosity = 1),
)
)
# ----------------------------------------
# Testing dataset
if 'testingFilename' in datasets:
datasetName = 'bothTesting'
inferSteps.append(
dict(name = '%s_baseline' % datasetName,
iterationCount = config['iterationCount'],
setup = [sensorOpen(datasets['testingFilename'])],
ppOptions = dict(printLearnedCoincidences=False),
)
)
inferSteps.append(
dict(name = '%s_acc' % datasetName,
iterationCount = config['iterationCount'],
setup = [sensorOpen(datasets['testingFilename'])],
ppOptions = dict(onlyClassificationAcc=True,
tpActivationThresholds=config['tpActivationThresholds']),
)
)
description['infer'] = inferSteps
return description
