def classifier_featureblock_logistic_regression(fm_train=traindat,
fm_test=testdat,
label_train=label_traindat):
from modshogun import BinaryLabels, RealFeatures, IndexBlock, IndexBlockGroup
try:
from modshogun import FeatureBlockLogisticRegression
except ImportError:
print("FeatureBlockLogisticRegression not available")
exit(0)
features = RealFeatures(hstack((traindat, traindat)))
labels = BinaryLabels(hstack((label_train, label_train)))
n_features = features.get_num_features()
block_one = IndexBlock(0, n_features // 2)
block_two = IndexBlock(n_features // 2, n_features)
block_group = IndexBlockGroup()
block_group.add_block(block_one)
block_group.add_block(block_two)
mtlr = FeatureBlockLogisticRegression(0.1, features, labels, block_group)
mtlr.set_regularization(1) # use regularization ratio
mtlr.set_tolerance(1e-2) # use 1e-2 tolerance
mtlr.train()
out = mtlr.apply().get_labels()
return out
def classifier_featureblock_logistic_regression (fm_train=traindat,fm_test=testdat,label_train=label_traindat):
from modshogun import BinaryLabels, RealFeatures, IndexBlock, IndexBlockGroup
try:
from modshogun import FeatureBlockLogisticRegression
except ImportError:
print("FeatureBlockLogisticRegression not available")
exit(0)
features = RealFeatures(hstack((traindat,traindat)))
labels = BinaryLabels(hstack((label_train,label_train)))
n_features = features.get_num_features()
block_one = IndexBlock(0,n_features//2)
block_two = IndexBlock(n_features//2,n_features)
block_group = IndexBlockGroup()
block_group.add_block(block_one)
block_group.add_block(block_two)
mtlr = FeatureBlockLogisticRegression(0.1,features,labels,block_group)
mtlr.set_regularization(1) # use regularization ratio
mtlr.set_tolerance(1e-2) # use 1e-2 tolerance
mtlr.train()
out = mtlr.apply().get_labels()
return out
def transfer_multitask_group_regression(fm_train=traindat,fm_test=testdat,label_train=label_traindat): from modshogun import RegressionLabels, RealFeatures, IndexBlock, IndexBlockGroup, MultitaskLSRegression features = RealFeatures(traindat) labels = RegressionLabels(label_train) n_vectors = features.get_num_vectors() task_one = IndexBlock(0,n_vectors/2) task_two = IndexBlock(n_vectors/2,n_vectors) task_group = IndexBlockGroup() task_group.add_block(task_one) task_group.add_block(task_two) mtlsr = MultitaskLSRegression(0.1,features,labels,task_group) mtlsr.train() mtlsr.set_current_task(0) out = mtlsr.apply_regression().get_labels() return out
num_z = 21
# get the data
X, y, true_coefs = generate_synthetic_logistic_data(
n, p, L, blk_nnz, gcov, nstd)
# here each column represents a feature vector
features = RealFeatures(X)
# we have to convert the labels to +1/-1
labels = BinaryLabels(np.sign(y.astype(int) - 0.5))
# SETTING UP THE CLASSIFIERS
# CLASSIFIER 1: group LASSO
# build the feature blocks and add them to the block group
pl = p / L
block_group = IndexBlockGroup()
for i in xrange(L):
block_group.add_block(IndexBlock(pl * i, pl * (i + 1)))
cls_gl = FeatureBlockLogisticRegression(0.0, features, labels, block_group)
# with set_regularization(1), the parameter z will indicate the fraction of
# the maximum regularization to use, and so z is in [0,1]
# (reference: SLEP manual)
cls_gl.set_regularization(1)
cls_gl.set_q(2.0) # it is the default anyway...
# CLASSIFIER 2: LASSO (illustrating group lasso with all group sizes = 1)
block_group_ones = IndexBlockGroup()
for i in xrange(p):
block_group_ones.add_block(IndexBlock(i, i + 1))
max_z = 1
num_z = 21
# get the data
X, y, true_coefs = generate_synthetic_logistic_data(n, p, L, blk_nnz, gcov, nstd)
# here each column represents a feature vector
features = RealFeatures(X)
# we have to convert the labels to +1/-1
labels = BinaryLabels(np.sign(y.astype(int) - 0.5))
# SETTING UP THE CLASSIFIERS
# CLASSIFIER 1: group LASSO
# build the feature blocks and add them to the block group
pl = p / L
block_group = IndexBlockGroup()
for i in xrange(L):
block_group.add_block(IndexBlock(pl * i, pl * (i + 1)))
cls_gl = FeatureBlockLogisticRegression(0.0, features, labels, block_group)
# with set_regularization(1), the parameter z will indicate the fraction of
# the maximum regularization to use, and so z is in [0,1]
# (reference: SLEP manual)
cls_gl.set_regularization(1)
cls_gl.set_q(2.0) # it is the default anyway...
# CLASSIFIER 2: LASSO (illustrating group lasso with all group sizes = 1)
block_group_ones = IndexBlockGroup()
for i in xrange(p):
block_group_ones.add_block(IndexBlock(i, i + 1))