def test_ANOVA_Refinement_Volume(self):
"""Dimensionally adaptive refinement using volume as local error
indicator
"""
# point ((3,7), (1,1)) (middle most right) gets larger surplus coefficient
alpha = DataVector(self.grid.getSize())
point_to_refine = None
for i in xrange(17):
point = self.grid_storage.get(i)
if point.getLevel(0) == 3 and point.getIndex(0) == 7 \
and point.getLevel(1) == 1 and point.getIndex(1) == 1:
point_to_refine = point
alpha[i] = 2.0
else:
alpha[i] = 1.0
# refine one point
functor = SurplusVolumeRefinementFunctor(alpha, 1, 0.0)
hash_refinement = HashRefinement()
refinement_strategy = ANOVARefinement(hash_refinement)
refinement_strategy.free_refine(self.grid_storage, functor)
# check if only the children along x1 direction were inserted
self.assertEqual(self.grid.getSize(), 19,
'Number of grid points doesn\'t match')
child = point_to_refine.__class__(point_to_refine)
self.grid_storage.left_child(child, 0)
self.assertTrue(self.grid_storage.has_key(child),
'Left x1 left child was not found')
child = point_to_refine.__class__(point_to_refine)
self.grid_storage.right_child(child, 0)
self.assertTrue(self.grid_storage.has_key(child),
'Left x1 right child was not found')
child = point_to_refine.__class__(point_to_refine)
self.grid_storage.left_child(child, 1)
self.assertFalse(
self.grid_storage.has_key(child),
'Left x2 left child is present, though should not be')
child = point_to_refine.__class__(point_to_refine)
self.grid_storage.right_child(child, 1)
self.assertFalse(
self.grid_storage.has_key(child),
'Left x2 right child is present, though should not be')
def test_ANOVA_Refinement_Volume(self):
"""Dimensionally adaptive refinement using volume as local error
indicator
"""
# point ((3,7), (1,1)) (middle most right) gets larger surplus coefficient
alpha = DataVector(self.grid.getSize())
point_to_refine = None
for i in xrange(17):
point = self.grid_storage.get(i)
if point.getLevel(0) == 3 and point.getIndex(0) == 7 \
and point.getLevel(1) == 1 and point.getIndex(1) == 1:
point_to_refine = point
alpha[i] = 2.0
else:
alpha[i] = 1.0
# refine one point
functor = SurplusVolumeRefinementFunctor(alpha, 1, 0.0)
hash_refinement = HashRefinement()
refinement_strategy = ANOVARefinement(hash_refinement)
refinement_strategy.free_refine(self.grid_storage, functor)
# check if only the children along x1 direction were inserted
self.assertEqual(self.grid.getSize(), 19,
'Number of grid points doesn\'t match')
child = point_to_refine.__class__(point_to_refine)
self.grid_storage.left_child(child, 0)
self.assertTrue(self.grid_storage.has_key(child),
'Left x1 left child was not found')
child = point_to_refine.__class__(point_to_refine)
self.grid_storage.right_child(child, 0)
self.assertTrue(self.grid_storage.has_key(child),
'Left x1 right child was not found')
child = point_to_refine.__class__(point_to_refine)
self.grid_storage.left_child(child, 1)
self.assertFalse(self.grid_storage.has_key(child),
'Left x2 left child is present, though should not be')
child = point_to_refine.__class__(point_to_refine)
self.grid_storage.right_child(child, 1)
self.assertFalse(self.grid_storage.has_key(child),
'Left x2 right child is present, though should not be')
def __testANOVAS(self, suffix):
# from bin.controller.InfoToScreen import InfoToScreen
builder = LearnerBuilder()
builder = builder.buildRegressor()
learner = builder.withTrainingDataFromCSVFile('refinement_strategy_%s.csv.gz'%suffix)\
.withGrid().withLevel(3)\
.withBorder(Types.BorderTypes.NONE)\
.withSpecification().withIdentityOperator().withAdaptThreshold(0.001)\
.withAdaptRate(1.0)\
.withLambda(0.0001)\
.withCGSolver().withImax(500)\
.withStopPolicy().withAdaptiveItarationLimit(5)\
.andGetResult()
learner.specification.setBOperator(
createOperationMultipleEval(
learner.grid,
learner.dataContainer.getPoints(DataContainer.TRAIN_CATEGORY)))
while True: #repeat until policy says "stop"
learner.notifyEventControllers(LearnerEvents.LEARNING_STEP_STARTED)
#learning step
learner.alpha = learner.doLearningIteration(learner.dataContainer)
learner.knowledge.update(learner.alpha)
#compress grid
if learner.iteration == 0:
generator = learner.grid.getGenerator()
functor = self.coarsening_functor(
learner.alpha, generator.getNumberOfRemovablePoints(),
learner.specification.getAdaptThreshold())
generator.coarsen(functor, learner.alpha)
self.plotGrid(learner, suffix)
storage = learner.grid.getStorage()
self.plot_grid_historgram(suffix, learner, storage)
formatter = GridImageFormatter()
formatter.serializeToFile(
learner.grid,
"%s%d_projections_anova.png" % (suffix, learner.iteration))
#calculate avg. error for training and test data and avg. for refine alpha
learner.updateResults(learner.alpha, learner.dataContainer)
learner.notifyEventControllers(
LearnerEvents.LEARNING_STEP_COMPLETE)
p_val = learner.trainAccuracy[-1] + learner.specification.getL(
) * np.sum(learner.alpha.array()**2)
print("ANOVA %s iteration %d: %d grid points, %1.9f MSE, p* = %1.10f" \
% (suffix, learner.iteration, storage.getSize(), learner.trainAccuracy[-1], p_val))
learner.iteration += 1
if learner.iteration == 5:
pass
if (learner.stopPolicy.isTrainingComplete(learner)): break
#refine grid
learner.notifyEventControllers(LearnerEvents.REFINING_GRID)
learner.grid.getStorage().recalcLeafProperty()
refinable_poits = learner.grid.getGenerator(
).getNumberOfRefinablePoints()
pointsNum = learner.specification.getNumOfPointsToRefine(
refinable_poits)
# learner.grid.getGenerator().refine( SurplusRefinementFunctor(learner.errors, int(pointsNum), learner.specification.getAdaptThreshold()) )
refiner = HashRefinement()
functor = self.refinement_functor(
learner.alpha, int(pointsNum),
learner.specification.getAdaptThreshold())
anova_refinement = ANOVARefinement(refiner)
anova_refinement.free_refine(learner.grid.getStorage(), functor)
#formatter = GridFormatter()
#formatter.serializeToFile(learner.grid, "grid_anova_%s.txt"%suffix)
del learner
def __testANOVAS(self, suffix):
# from bin.controller.InfoToScreen import InfoToScreen
builder = LearnerBuilder()
builder = builder.buildRegressor()
learner = builder.withTrainingDataFromCSVFile('refinement_strategy_%s.csv.gz'%suffix)\
.withGrid().withLevel(3)\
.withBorder(Types.BorderTypes.NONE)\
.withSpecification().withIdentityOperator().withAdaptThreshold(0.001)\
.withAdaptRate(1.0)\
.withLambda(0.0001)\
.withCGSolver().withImax(500)\
.withStopPolicy().withAdaptiveItarationLimit(5)\
.andGetResult()
learner.specification.setBOperator(createOperationMultipleEval(learner.grid,
learner.dataContainer.getPoints(DataContainer.TRAIN_CATEGORY)))
while True: #repeat until policy says "stop"
learner.notifyEventControllers(LearnerEvents.LEARNING_STEP_STARTED)
#learning step
learner.alpha = learner.doLearningIteration(learner.dataContainer)
learner.knowledge.update(learner.alpha)
#compress grid
if learner.iteration == 0:
generator = learner.grid.createGridGenerator()
functor = self.coarsening_functor(
learner.alpha,
generator.getNumberOfRemovablePoints(),
0.99, learner.grid.getStorage())
# functor = self.coarsening_functor(
# learner.alpha,
# generator.getNumberOfRemovablePoints(),
# learner.specification.getAdaptThreshold())
generator.coarsen(functor, learner.alpha)
#print "coersening finished"
self.plotGrid(learner, suffix)
storage = learner.grid.getStorage()
self.plot_grid_historgram(suffix, learner, storage)
formatter = GridImageFormatter()
formatter.serializeToFile(learner.grid, "%s%d_projections_anova.png"%(suffix, learner.iteration))
#calculate avg. error for training and test data and avg. for refine alpha
learner.updateResults(learner.alpha, learner.dataContainer)
learner.notifyEventControllers(LearnerEvents.LEARNING_STEP_COMPLETE)
p_val = learner.trainAccuracy[-1] + learner.specification.getL()*np.sum(learner.alpha.array()**2)
print "ANOVA %s iteration %d: %d grid points, %1.9f MSE, p* = %1.10f" % \
(suffix, learner.iteration, storage.size(), learner.trainAccuracy[-1], p_val)
learner.iteration += 1
if learner.iteration == 5:
pass
if(learner.stopPolicy.isTrainingComplete(learner)): break
#refine grid
learner.notifyEventControllers(LearnerEvents.REFINING_GRID)
learner.grid.getStorage().recalcLeafProperty()
refinable_poits = learner.grid.createGridGenerator().getNumberOfRefinablePoints()
pointsNum = learner.specification.getNumOfPointsToRefine(refinable_poits)
# learner.grid.createGridGenerator().refine( SurplusRefinementFunctor(learner.errors, int(pointsNum), learner.specification.getAdaptThreshold()) )
refiner = HashRefinement()
functor = self.refinement_functor(learner.alpha, int(pointsNum), learner.specification.getAdaptThreshold())
anova_refinement = ANOVARefinement(refiner)
anova_refinement.free_refine(learner.grid.getStorage(), functor)
#formatter = GridFormatter()
#formatter.serializeToFile(learner.grid, "grid_anova_%s.txt"%suffix)
del learner