def __init__(self,
minValue,
maxValue,
baseThreshold=0.75,
restPeriod=30,
maxLeftSemiContextsLenght=7,
maxActiveNeuronsNum=15,
numNormValueBits=3):
self.minValue = float(minValue)
self.maxValue = float(maxValue)
self.restPeriod = restPeriod
self.baseThreshold = baseThreshold
self.maxActNeurons = maxActiveNeuronsNum
self.numNormValueBits = numNormValueBits
self.maxBinValue = 2**self.numNormValueBits - 1.0
self.fullValueRange = self.maxValue - self.minValue
if self.fullValueRange == 0.0:
self.fullValueRange = self.maxBinValue
self.minValueStep = self.fullValueRange / self.maxBinValue
self.leftFactsGroup = tuple()
self.contextOperator = ContextOperator(maxLeftSemiContextsLenght)
self.potentialNewContexts = []
self.aScoresHistory = [1.0]
def __init__( self,
minValue,
maxValue,
baseThreshold = 0.75,
restPeriod = 30,
maxLeftSemiContextsLenght = 7,
maxActiveNeuronsNum = 15,
numNormValueBits = 3 ) :
self.minValue = float(minValue)
self.maxValue = float(maxValue)
self.restPeriod = restPeriod
self.baseThreshold = baseThreshold
self.maxActNeurons = maxActiveNeuronsNum
self.numNormValueBits = numNormValueBits
self.maxBinValue = 2 ** self.numNormValueBits - 1.0
self.fullValueRange = self.maxValue - self.minValue
if self.fullValueRange == 0.0 :
self.fullValueRange = self.maxBinValue
self.minValueStep = self.fullValueRange / self.maxBinValue
self.leftFactsGroup = tuple()
self.contextOperator = ContextOperator( maxLeftSemiContextsLenght )
self.potentialNewContexts = []
self.aScoresHistory = [ 1.0 ]
class ContextualAnomalyDetectorOSE(object):
"""
Contextual Anomaly Detector - Open Source Edition
2016, Mikhail Smirnov [email protected]
https://github.com/smirmik/CAD
"""
def __init__(self,
minValue,
maxValue,
baseThreshold=0.75,
restPeriod=30,
maxLeftSemiContextsLenght=7,
maxActiveNeuronsNum=15,
numNormValueBits=3):
self.minValue = float(minValue)
self.maxValue = float(maxValue)
self.restPeriod = restPeriod
self.baseThreshold = baseThreshold
self.maxActNeurons = maxActiveNeuronsNum
self.numNormValueBits = numNormValueBits
self.maxBinValue = 2**self.numNormValueBits - 1.0
self.fullValueRange = self.maxValue - self.minValue
if self.fullValueRange == 0.0:
self.fullValueRange = self.maxBinValue
self.minValueStep = self.fullValueRange / self.maxBinValue
self.leftFactsGroup = tuple()
self.contextOperator = ContextOperator(maxLeftSemiContextsLenght)
self.potentialNewContexts = []
self.aScoresHistory = [1.0]
def step(self, inpFacts):
currSensFacts = tuple(sorted(set(inpFacts)))
uniqPotNewContexts = set()
if len(self.leftFactsGroup) > 0 and len(currSensFacts) > 0:
potNewZeroLevelContext = tuple(
[self.leftFactsGroup, currSensFacts])
uniqPotNewContexts.add(potNewZeroLevelContext)
newContextFlag = self.contextOperator.getContextByFacts(
[potNewZeroLevelContext], zerolevel=1)
else:
newContextFlag = False
leftCrossing = self.contextOperator.contextCrosser(
leftOrRight=1,
factsList=currSensFacts,
newContextFlag=newContextFlag)
activeContexts, numSelContexts, potNewContexts = leftCrossing
uniqPotNewContexts.update(potNewContexts)
numUniqPotNewContext = len(uniqPotNewContexts)
if numSelContexts > 0:
percentSelectedContextActive = len(activeContexts) / float(
numSelContexts)
else:
percentSelectedContextActive = 0.0
srtAContexts = sorted(activeContexts, cmp=aContextsCMP)
activeNeurons = [
cInf[0] for cInf in srtAContexts[-self.maxActNeurons:]
]
currNeurFacts = set([2**31 + fact for fact in activeNeurons])
leftFactsGroup = set()
leftFactsGroup.update(currSensFacts, currNeurFacts)
self.leftFactsGroup = tuple(sorted(leftFactsGroup))
numNewCont = self.contextOperator.contextCrosser(
leftOrRight=0,
factsList=self.leftFactsGroup,
potentialNewContexts=potNewContexts)
numNewCont += 1 if newContextFlag else 0
if newContextFlag and numUniqPotNewContext > 0:
percentAddedContextToUniqPotNew = numNewCont / float(
numUniqPotNewContext)
else:
percentAddedContextToUniqPotNew = 0.0
return percentSelectedContextActive, percentAddedContextToUniqPotNew
def getAnomalyScore(self, inputData):
normInpVal = int(
(inputData["value"] - self.minValue) / self.minValueStep)
binInpValue = bin(normInpVal).lstrip("0b").rjust(
self.numNormValueBits, "0")
outSens = []
for sNum, currSymb in enumerate(reversed(binInpValue)):
outSens.append(sNum * 2 + (1 if currSymb == "1" else 0))
setOutSens = set(outSens)
anomalyVal1, anomalyVal2 = self.step(setOutSens)
currentAnomalyScore = (1.0 - anomalyVal1 + anomalyVal2) / 2.0
if max(self.aScoresHistory[-int(self.restPeriod):]
) < self.baseThreshold:
returnedAnomalyScore = currentAnomalyScore
else:
returnedAnomalyScore = 0.0
self.aScoresHistory.append(currentAnomalyScore)
return returnedAnomalyScore
class ContextualAnomalyDetectorOSE(object): """ Contextual Anomaly Detector - Open Source Edition 2016, Mikhail Smirnov [email protected] https://github.com/smirmik/CAD """ def __init__( self, minValue, maxValue, baseThreshold = 0.75, restPeriod = 30, maxLeftSemiContextsLenght = 7, maxActiveNeuronsNum = 15, numNormValueBits = 3 ) : self.minValue = float(minValue) self.maxValue = float(maxValue) self.restPeriod = restPeriod self.baseThreshold = baseThreshold self.maxActNeurons = maxActiveNeuronsNum self.numNormValueBits = numNormValueBits self.maxBinValue = 2 ** self.numNormValueBits - 1.0 self.fullValueRange = self.maxValue - self.minValue if self.fullValueRange == 0.0 : self.fullValueRange = self.maxBinValue self.minValueStep = self.fullValueRange / self.maxBinValue self.leftFactsGroup = tuple() self.contextOperator = ContextOperator( maxLeftSemiContextsLenght ) self.potentialNewContexts = [] self.aScoresHistory = [ 1.0 ] def step(self, inpFacts): currSensFacts = tuple(sorted(set(inpFacts))) uniqPotNewContexts = set() if len(self.leftFactsGroup) > 0 and len(currSensFacts) > 0 : potNewZeroLevelContext = tuple([self.leftFactsGroup,currSensFacts]) uniqPotNewContexts.add(potNewZeroLevelContext) newContextFlag = self.contextOperator.getContextByFacts( [potNewZeroLevelContext], zerolevel = 1 ) else : newContextFlag = False leftCrossing = self.contextOperator.contextCrosser ( leftOrRight = 1, factsList = currSensFacts, newContextFlag = newContextFlag ) activeContexts, numSelContexts, potNewContexts = leftCrossing uniqPotNewContexts.update(potNewContexts) numUniqPotNewContext = len(uniqPotNewContexts) if numSelContexts > 0 : percentSelectedContextActive = len(activeContexts) / float(numSelContexts) else : percentSelectedContextActive = 0.0 srtAContexts = sorted(activeContexts, cmp=aContextsCMP) activeNeurons = [ cInf[0] for cInf in srtAContexts[-self.maxActNeurons:] ] currNeurFacts = set([ 2 ** 31 + fact for fact in activeNeurons ]) leftFactsGroup = set() leftFactsGroup.update(currSensFacts, currNeurFacts) self.leftFactsGroup = tuple(sorted(leftFactsGroup)) numNewCont = self.contextOperator.contextCrosser ( leftOrRight = 0, factsList = self.leftFactsGroup, potentialNewContexts = potNewContexts ) numNewCont += 1 if newContextFlag else 0 if newContextFlag and numUniqPotNewContext > 0 : percentAddedContextToUniqPotNew = numNewCont / float(numUniqPotNewContext) else : percentAddedContextToUniqPotNew = 0.0 return percentSelectedContextActive, percentAddedContextToUniqPotNew def getAnomalyScore(self,inputData): normInpVal = int((inputData["value"] - self.minValue) / self.minValueStep) binInpValue = bin(normInpVal).lstrip("0b").rjust(self.numNormValueBits,"0") outSens = [] for sNum, currSymb in enumerate(reversed(binInpValue)) : outSens.append( sNum * 2 + ( 1 if currSymb == "1" else 0 ) ) setOutSens = set(outSens) anomalyVal1, anomalyVal2 = self.step(setOutSens) currentAnomalyScore = (1.0 - anomalyVal1 + anomalyVal2) / 2.0 if max(self.aScoresHistory[-int(self.restPeriod):]) < self.baseThreshold : returnedAnomalyScore = currentAnomalyScore else : returnedAnomalyScore = 0.0 self.aScoresHistory.append(currentAnomalyScore) return returnedAnomalyScore
