diff --git a/Code/MonoMutliViewClassifiers/Monoview/ExecClassifMonoView.py b/Code/MonoMutliViewClassifiers/Monoview/ExecClassifMonoView.py
index 67b86d4033c9016fef4d18ee1a7c308a0b2da503..92653a2ff7b4c04e67019cbf070b0fc813d76996 100644
--- a/Code/MonoMutliViewClassifiers/Monoview/ExecClassifMonoView.py
+++ b/Code/MonoMutliViewClassifiers/Monoview/ExecClassifMonoView.py
@@ -70,7 +70,7 @@ def ExecMonoview(X, Y, name, labelsNames, learningRate, nbFolds, nbCores, databa
y_test_preds = []
for iterationStat in range(statsIter):
# Calculate Train/Test data
- logging.debug("Start:\t Determine Train/Test split")
+ logging.debug("Start:\t Determine Train/Test split"+" for iteration "+str(iterationStat+1))
testIndices = ClassifMonoView.splitDataset(Y, nbClass, learningRate, datasetLength)
trainIndices = [i for i in range(datasetLength) if i not in testIndices]
diff --git a/Code/MonoMutliViewClassifiers/Multiview/ExecMultiview.py b/Code/MonoMutliViewClassifiers/Multiview/ExecMultiview.py
index 8f2aa96e03e87957244ed5616964bff6d791e400..77769d796b0e1e1a5946630793d416368d9b2625 100644
--- a/Code/MonoMutliViewClassifiers/Multiview/ExecMultiview.py
+++ b/Code/MonoMutliViewClassifiers/Multiview/ExecMultiview.py
@@ -68,9 +68,10 @@ def ExecMultiview(DATASET, name, learningRate, nbFolds, nbCores, databaseType, p
classifierClass = getattr(classifierModule, CL_type)
classifierGridSearch = getattr(classifierModule, "gridSearch_hdf5")
analysisModule = getattr(classifierPackage, "analyzeResults")
+
for iterIndex in range(statsIter):
- logging.info("Start:\t Determine validation split for ratio " + str(learningRate))
+ logging.info("Start:\t Determine validation split for ratio " + str(learningRate)+" for iteration "+str(iterIndex+1))
validationIndices = DB.splitDataset(DATASET, learningRate, datasetLength)
learningIndices = [index for index in range(datasetLength) if index not in validationIndices]
classificationSetLength = len(learningIndices)
@@ -88,23 +89,6 @@ def ExecMultiview(DATASET, name, learningRate, nbFolds, nbCores, databaseType, p
logging.info("Start:\t Learning with " + CL_type + " and " + str(len(kFolds)) + " folds")
-
-
-
-
-
-
- if gridSearch:
- logging.info("Start:\t Randomsearching best settings for monoview classifiers")
- bestSettings, fusionConfig = classifierGridSearch(DATASET, viewsIndices, classificationKWARGS, learningIndices
- , metric=metrics[0], nIter=nIter)
- classificationKWARGS["classifiersConfigs"] = bestSettings
- try:
- classificationKWARGS["fusionMethodConfig"] = fusionConfig
- except:
- pass
- logging.info("Done:\t Randomsearching best settings for monoview classifiers")
-
logging.info("Start:\t Classification")
# Begin Classification
@@ -119,6 +103,16 @@ def ExecMultiview(DATASET, name, learningRate, nbFolds, nbCores, databaseType, p
fold.sort()
logging.info("\tStart:\t Fold number " + str(foldIdx + 1))
trainIndices = [index for index in range(datasetLength) if (index not in fold) and (index not in validationIndices)]
+ if gridSearch:
+ logging.info("Start:\t Randomsearching best settings for monoview classifiers")
+ bestSettings, fusionConfig = classifierGridSearch(DATASET, viewsIndices, classificationKWARGS, trainIndices
+ , metric=metrics[0], nIter=nIter)
+ classificationKWARGS["classifiersConfigs"] = bestSettings
+ try:
+ classificationKWARGS["fusionMethodConfig"] = fusionConfig
+ except:
+ pass
+ logging.info("Done:\t Randomsearching best settings for monoview classifiers")
DATASET_LENGTH = len(trainIndices)
classifier = classifierClass(NB_VIEW, DATASET_LENGTH, DATASET.get("Labels").value[trainIndices], NB_CORES=nbCores, **classificationKWARGS)
@@ -127,7 +121,6 @@ def ExecMultiview(DATASET, name, learningRate, nbFolds, nbCores, databaseType, p
learningTime = time.time() - extractionTime - t_start
kFoldLearningTimeIter.append(learningTime)
- logging.info("\tStart: \t Classification")
kFoldPredictedTrainLabelsIter.append(classifier.predict_hdf5(DATASET, usedIndices=trainIndices, viewsIndices=viewsIndices))
kFoldPredictedTestLabelsIter.append(classifier.predict_hdf5(DATASET, usedIndices=fold, viewsIndices=viewsIndices))
kFoldPredictedValidationLabelsIter.append(classifier.predict_hdf5(DATASET, usedIndices=validationIndices, viewsIndices=viewsIndices))
diff --git a/Code/MonoMutliViewClassifiers/Multiview/Fusion/Methods/LateFusionPackage/SCMForLinear.py b/Code/MonoMutliViewClassifiers/Multiview/Fusion/Methods/LateFusionPackage/SCMForLinear.py
index 57888231b715addfed55322a7c20e3811cc8c810..6d9bc56403a6c4978bf731b0ac29aac4a7c05b0a 100644
--- a/Code/MonoMutliViewClassifiers/Multiview/Fusion/Methods/LateFusionPackage/SCMForLinear.py
+++ b/Code/MonoMutliViewClassifiers/Multiview/Fusion/Methods/LateFusionPackage/SCMForLinear.py
@@ -68,11 +68,15 @@ class SCMForLinear(LateFusionClassifier):
usedIndices = range(DATASET.get("Metadata").attrs["datasetLength"])
if usedIndices:
monoviewDecisions = np.zeros((len(usedIndices), nbView), dtype=int)
+ accus=[]
for index, viewIndex in enumerate(viewsIndices):
- monoviewDecisions[:, index] = self.monoviewClassifiers[index].predict(
+ monoviewDecision = self.monoviewClassifiers[index].predict(
getV(DATASET, viewIndex, usedIndices))
- features = self.generateInteractions(monoviewDecisions, order=self.order)
+ accus.append(accuracy_score(DATASET.get("Labels").value[usedIndices], monoviewDecision))
+ monoviewDecisions[:, index] = monoviewDecision
+ features = self.generateInteractions(monoviewDecisions)
predictedLabels = self.SCMClassifier.predict(features)
+ print str(np.array([accuracy_score(DATASET.get("Labels").value[usedIndices], predictedLabels)>acc for acc in accus]).all()), len(self.SCMClassifier.model)
else:
predictedLabels = []
return predictedLabels
@@ -91,8 +95,10 @@ class SCMForLinear(LateFusionClassifier):
for index, viewIndex in enumerate(viewsIndices):
monoViewDecisions[:, index] = self.monoviewClassifiers[index].predict(
getV(DATASET, viewIndex, usedIndices))
- features = self.generateInteractions(monoViewDecisions, order=self.order)
- featureSequence = [str(featureIndex) for featureIndex in range(features.shape[1])]
+ features = self.generateInteractions(monoViewDecisions)
+ featureSequence=[str(index) for index in range(nbView)]
+ for orderIndex in range(self.order-1):
+ featureSequence += [str(featureIndex) for featureIndex in itertools.combinations(range(monoViewDecisions.shape[1]), orderIndex+2)]
featureIndexByRule = np.arange(features.shape[1], dtype=np.uint32)
binaryAttributes = LazyBaptisteRuleList(featureSequence, featureIndexByRule)
packedData = _pack_binary_bytes_to_ints(features, 64)
@@ -124,14 +130,14 @@ class SCMForLinear(LateFusionClassifier):
except:
pass
- def generateInteractions(self, monoViewDecisions, order=None):
- if type(order)==type(None):
+ def generateInteractions(self, monoViewDecisions):
+ if type(self.order)==type(None):
order = monoViewDecisions.shape[1]
genratedIntercations = [monoViewDecisions[:,i] for i in range(monoViewDecisions.shape[1])]
- if order==1:
+ if self.order==1:
return monoViewDecisions
else:
- for orderIndex in range(order-1):
+ for orderIndex in range(self.order-1):
combins = itertools.combinations(range(monoViewDecisions.shape[1]), orderIndex+2)
for combin in combins:
generatedDecision = monoViewDecisions[:,combin[0]]
diff --git a/Code/MonoMutliViewClassifiers/Multiview/GetMultiviewDb.py b/Code/MonoMutliViewClassifiers/Multiview/GetMultiviewDb.py
index 9d1ec462239af1467ac4907a10fff5b2e29c764a..34257b61b5ac1fd3f7755091e6d6ab6d50777d54 100644
--- a/Code/MonoMutliViewClassifiers/Multiview/GetMultiviewDb.py
+++ b/Code/MonoMutliViewClassifiers/Multiview/GetMultiviewDb.py
@@ -13,18 +13,33 @@ __author__ = "Baptiste Bauvin"
__status__ = "Prototype" # Production, Development, Prototype
-def getPlausibleDBhdf5(features, pathF, name , NB_CLASS, LABELS_NAME, nbView=4, nbClass=2, datasetLength=500):
+def makeMeNoisy(viewData, percentage=5):
+ viewData = viewData.astype(bool)
+ nbNoisyCoord = int(percentage/100.0*viewData.shape[0]*viewData.shape[1])
+ rows = range(viewData.shape[0])
+ cols = range(viewData.shape[1])
+ for _ in range(nbNoisyCoord):
+ rowIdx = random.choice(rows)
+ colIdx = random.choice(cols)
+ viewData[rowIdx, colIdx] = not viewData[rowIdx, colIdx]
+ noisyViewData = viewData.astype(np.uint8)
+ return noisyViewData
+
+
+def getPlausibleDBhdf5(features, pathF, name , NB_CLASS, LABELS_NAME, nbView=10, nbClass=2, datasetLength=500):
nbFeatures = 150
datasetFile = h5py.File(pathF+"Plausible.hdf5", "w")
CLASS_LABELS = np.array([0 for i in range(datasetLength/2)]+[1 for i in range(datasetLength/2)])
for viewIndex in range(nbView):
# if viewIndex== 0 :
- viewData = np.array([np.zeros(nbFeatures) for i in range(datasetLength/2)]+[np.ones(nbFeatures) for i in range(datasetLength/2)]).astype(np.uint8)
- fakeTrueIndices = np.random.randint(0, datasetLength/2-1, datasetLength/5)
- fakeFalseIndices = np.random.randint(datasetLength/2, datasetLength, datasetLength/5)
+ viewData = np.array([np.zeros(nbFeatures) for i in range(datasetLength/2)]+[np.ones(nbFeatures) for i in range(datasetLength/2)])
+ fakeTrueIndices = np.random.randint(0, datasetLength/2-1, datasetLength/10)
+ fakeFalseIndices = np.random.randint(datasetLength/2, datasetLength, datasetLength/10)
+
viewData[fakeTrueIndices] = np.ones((len(fakeTrueIndices), nbFeatures))
viewData[fakeFalseIndices] = np.zeros((len(fakeFalseIndices), nbFeatures))
- viewDset = datasetFile.create_dataset("View"+str(viewIndex), viewData.shape, data=viewData)
+ viewData = makeMeNoisy(viewData)
+ viewDset = datasetFile.create_dataset("View"+str(viewIndex), viewData.shape, data=viewData.astype(np.uint8))
viewDset.attrs["name"] = "View"+str(viewIndex)
viewDset.attrs["sparse"] = False
viewDset.attrs["binary"] = True
@@ -49,7 +64,7 @@ def getPlausibleDBhdf5(features, pathF, name , NB_CLASS, LABELS_NAME, nbView=4,
def getFakeDBhdf5(features, pathF, name , NB_CLASS, LABELS_NAME):
NB_VIEW = 4
- DATASET_LENGTH = 300
+ DATASET_LENGTH = 30
NB_CLASS = 2
VIEW_DIMENSIONS = np.random.random_integers(5, 20, NB_VIEW)
diff --git a/Code/MonoMutliViewClassifiers/ResultAnalysis.py b/Code/MonoMutliViewClassifiers/ResultAnalysis.py
index 951c9f968b7b191dd8883bc4861c2abde8f6b615..1f80cf1ef4022c4828f123a845fb03d0e48aedfa 100644
--- a/Code/MonoMutliViewClassifiers/ResultAnalysis.py
+++ b/Code/MonoMutliViewClassifiers/ResultAnalysis.py
@@ -48,7 +48,7 @@ def resultAnalysis(benchmark, results, name, times, metrics):
metricKWARGS = {}
ax.set_title(getattr(Metrics, metric[0]).getConfig(**metricKWARGS)+" on validation set for each classifier")
rects = ax.bar(range(nbResults), validationScores, width, color="r")
- rect2 = ax.bar(np.arange(nbResults)+width, trainScores, width, color="0.3")
+ rect2 = ax.bar(np.arange(nbResults)+width, trainScores, width, color="0.7")
autolabel(rects, ax)
autolabel(rect2, ax)
ax.legend((rects[0], rect2[0]), ('Train', 'Test'))
diff --git a/Code/MonoMutliViewClassifiers/Versions.py b/Code/MonoMutliViewClassifiers/Versions.py
index ccff8b4a2c4bad8d7dc47c9326e43081cf757b4c..3f0038a3b22b70ba489a9405f188b1fd143095d8 100644
--- a/Code/MonoMutliViewClassifiers/Versions.py
+++ b/Code/MonoMutliViewClassifiers/Versions.py
@@ -21,6 +21,11 @@ def testVersions():
except:
print "Please install Python 2.7"
raise
+ try:
+ import pyscm
+ except:
+ print "Please install pyscm"
+ raise
try:
import cv2