def prepare_data(use_toy=True):
from os.path import exists
from tools.load import LoadMatrix
lm=LoadMatrix()
if not use_toy and exists('../data/../mldata/uci-20070111-optdigits.mat'):
from scipy.io import loadmat
mat = loadmat('../data/../mldata/uci-20070111-optdigits.mat')['int0'].astype(float)
X = mat[:-1,:]
Y = mat[-1,:]
isplit = X.shape[1]/2
traindat = X[:,:isplit]
label_traindat = Y[:isplit]
testdat = X[:, isplit:]
label_testdat = Y[isplit:]
else:
traindat = lm.load_numbers('../data/fm_train_real.dat')
testdat = lm.load_numbers('../data/fm_test_real.dat')
label_traindat = lm.load_labels('../data/label_train_multiclass.dat')
label_testdat = None
return [traindat, label_traindat, testdat, label_testdat]
# In this example the Histogram algorithm object computes a histogram over all
# 16bit unsigned integers in the features.
from tools.load import LoadMatrix
from sg import sg
lm=LoadMatrix()
traindna=lm.load_dna('../data/fm_train_dna.dat')
cubedna=lm.load_cubes('../data/fm_train_cube.dat')
parameter_list=[[traindna,cubedna,3,0,'n'],[traindna,cubedna,4,0,'n']]
def distribution_histogram(fm_train=traindna,fm_cube=cubedna,order=3,
gap=0,reverse='n'):
# sg('new_distribution', 'HISTOGRAM')
sg('add_preproc', 'SORTWORDSTRING')
sg('set_features', 'TRAIN', fm_train, 'DNA')
sg('convert', 'TRAIN', 'STRING', 'CHAR', 'STRING', 'WORD', order, order-1, gap, reverse)
sg('attach_preproc', 'TRAIN')
# sg('train_distribution')
# histo=sg('get_histogram')
# num_examples=11
# num_param=sg('get_histogram_num_model_parameters')
# for i in xrange(num_examples):
# for j in xrange(num_param):
# sg('get_log_derivative %d %d' % (j, i))
# sg('get_log_likelihood')
# return sg('get_log_likelihood_sample')
###########################################################################
# linear kernel on byte features
###########################################################################
from tools.load import LoadMatrix
from numpy import ubyte
lm=LoadMatrix()
traindat = ubyte(lm.load_numbers('../data/fm_train_byte.dat'))
testdat = ubyte(lm.load_numbers('../data/fm_test_byte.dat'))
parameter_list=[[traindat,testdat],[traindat,testdat]]
def kernel_linear_byte_modular(fm_train_byte=traindat,fm_test_byte=testdat):
from shogun.Kernel import LinearKernel
from shogun.Features import ByteFeatures
feats_train=ByteFeatures(fm_train_byte)
feats_test=ByteFeatures(fm_test_byte)
kernel=LinearKernel(feats_train, feats_train)
km_train=kernel.get_kernel_matrix()
kernel.init(feats_train, feats_test)
km_test=kernel.get_kernel_matrix()
return kernel
if __name__=='__main__':
print('LinearByte')
kernel_linear_byte_modular(*parameter_list[0])
from tools.load import LoadMatrix
from numpy import random
lm=LoadMatrix()
ground_truth = lm.load_labels('../data/label_train_twoclass.dat')
random.seed(17)
predicted = random.randn(len(ground_truth))
parameter_list = [[ground_truth,predicted]]
def evaluation_prcevaluation_modular(ground_truth, predicted):
from shogun.Features import Labels
from shogun.Evaluation import PRCEvaluation
ground_truth_labels = Labels(ground_truth)
predicted_labels = Labels(predicted)
evaluator = PRCEvaluation()
evaluator.evaluate(predicted_labels,ground_truth_labels)
return evaluator.get_PRC(), evaluator.get_auPRC()
if __name__=='__main__':
print 'PRCEvaluation'
evaluation_prcevaluation_modular(*parameter_list[0])
reverse=False
charfeat=StringCharFeatures(DNA)
charfeat.set_features(fm_dna)
feats=StringWordFeatures(charfeat.get_alphabet())
feats.obtain_from_char(charfeat, order-1, order, gap, reverse)
histo=Histogram(feats)
histo.train()
histo.get_histogram()
num_examples=feats.get_num_vectors()
num_param=histo.get_num_model_parameters()
#for i in xrange(num_examples):
# for j in xrange(num_param):
# histo.get_log_derivative(j, i)
histo.get_log_likelihood()
histo.get_log_likelihood_sample()
###########################################################################
# call functions
###########################################################################
if __name__=='__main__':
from tools.load import LoadMatrix
lm=LoadMatrix()
fm_dna=lm.load_dna('../data/fm_train_dna.dat')
histogram()
from tools.load import LoadMatrix
lm=LoadMatrix()
traindna = lm.load_dna('../data/fm_train_dna.dat')
testdna = lm.load_dna('../data/fm_test_dna.dat')
testdat = lm.load_numbers('../data/fm_test_real.dat')
parameter_list = [[traindna,testdna,testdat,3,0,False],[traindna,testdna,testdat,4,0,False]]
def distance_manhattenword_modular (fm_train_dna=traindna ,fm_test_dna=testdna,fm_test_real=testdat,order=3,gap=0,reverse=False):
from shogun.Features import StringCharFeatures, StringWordFeatures, DNA
from shogun.Preprocessor import SortWordString
from shogun.Distance import ManhattanWordDistance
charfeat=StringCharFeatures(DNA)
charfeat.set_features(fm_train_dna)
feats_train=StringWordFeatures(charfeat.get_alphabet())
feats_train.obtain_from_char(charfeat, order-1, order, gap, reverse)
preproc=SortWordString()
preproc.init(feats_train)
feats_train.add_preprocessor(preproc)
feats_train.apply_preprocessor()
charfeat=StringCharFeatures(DNA)
charfeat.set_features(fm_test_dna)
feats_test=StringWordFeatures(charfeat.get_alphabet())
feats_test.obtain_from_char(charfeat, order-1, order, gap, reverse)
feats_test.add_preprocessor(preproc)
feats_test.apply_preprocessor()
from tools.load import LoadMatrix
from sg import sg
lm=LoadMatrix()
traindat=lm.load_numbers('../data/fm_train_real.dat')
testdat=lm.load_numbers('../data/fm_test_real.dat')
parameter_list=[[traindat,testdat,1.4,10,True],[traindat,testdat,1.5,11,True]]
def preproc_prunevarsubmean (fm_train_real=traindat,fm_test_real=testdat,
width=1.4,size_cache=10,divide_by_std=True):
sg('add_preproc', 'PRUNEVARSUBMEAN', divide_by_std)
sg('set_kernel', 'CHI2', 'REAL', size_cache, width)
sg('set_features', 'TRAIN', fm_train_real)
sg('attach_preproc', 'TRAIN')
km=sg('get_kernel_matrix', 'TRAIN')
sg('set_features', 'TEST', fm_test_real)
sg('attach_preproc', 'TEST')
km=sg('get_kernel_matrix', 'TEST')
return km
if __name__=='__main__':
print 'PruneVarSubMean'
preproc_prunevarsubmean(*parameter_list[0])
# In this example a two-class support vector machine classifier is trained on a
# DNA splice-site detection data set and the trained classifier is used to predict
# labels on test set. As training algorithm SVM^light is used with SVM
# regularization parameter C=1.2 and the Weighted Degree kernel of degree 20 and
# the precision parameter epsilon=1e-5.
#
# For more details on the SVM^light see
# T. Joachims. Making large-scale SVM learning practical. In Advances in Kernel
# Methods -- Support Vector Learning, pages 169-184. MIT Press, Cambridge, MA USA, 1999.
#
# For more details on the Weighted Degree kernel see
# G. Raetsch, S.Sonnenburg, and B. Schoelkopf. RASE: recognition of alternatively
# spliced exons in C. elegans. Bioinformatics, 21:369-377, June 2005.
from tools.load import LoadMatrix
lm=LoadMatrix()
traindat = lm.load_dna('../data/fm_train_dna.dat')
testdat = lm.load_dna('../data/fm_test_dna.dat')
label_traindat = lm.load_labels('../data/label_train_dna.dat')
parameter_list = [[traindat,testdat,label_traindat,1.1,1e-5,1],[traindat,testdat,label_traindat,1.2,1e-5,1]]
def classifier_svmlight_modular (fm_train_dna=traindat,fm_test_dna=testdat,label_train_dna=label_traindat,C=1.2,epsilon=1e-5,num_threads=1):
from shogun.Features import StringCharFeatures, Labels, DNA
from shogun.Kernel import WeightedDegreeStringKernel
try:
from shogun.Classifier import SVMLight
except ImportError:
print 'No support for SVMLight available.'
return
#!/usr/bin/env python
from tools.load import LoadMatrix
from numpy import random
lm = LoadMatrix()
random.seed(17)
from tools.multiclass_shared import prepare_data
[traindat, label_traindat, testdat, label_testdat] = prepare_data(False)
parameter_list = [[traindat, label_traindat, testdat, label_testdat]]
def evaluation_multiclassovrevaluation_modular(traindat, label_traindat,
testdat, label_testdat):
from shogun.Features import MulticlassLabels
from shogun.Evaluation import MulticlassOVREvaluation, ROCEvaluation
from modshogun import MulticlassLibLinear, RealFeatures, ContingencyTableEvaluation, ACCURACY
from shogun.Mathematics import Math
Math.init_random(1)
ground_truth_labels = MulticlassLabels(label_traindat)
svm = MulticlassLibLinear(1.0, RealFeatures(traindat),
MulticlassLabels(label_traindat))
svm.train()
predicted_labels = svm.apply()
binary_evaluator = ROCEvaluation()
evaluator = MulticlassOVREvaluation(binary_evaluator)
mean_roc = evaluator.evaluate(predicted_labels, ground_truth_labels)
#print mean_roc
from tools.load import LoadMatrix
lm = LoadMatrix()
data = lm.load_numbers('../data/fm_train_real.dat')
parameter_list = [[data, 10], [data, 20]]
def converter_localtangentspacealignment_modular(data, k):
from shogun.Features import RealFeatures
from shogun.Converter import LocalTangentSpaceAlignment
features = RealFeatures(data)
converter = LocalTangentSpaceAlignment()
converter.set_target_dim(1)
converter.set_k(k)
converter.apply(features)
return features
if __name__ == '__main__':
print 'LocalTangentSpaceAlignment'
converter_localtangentspacealignment_modular(*parameter_list[0])
from tools.load import LoadMatrix
lm = LoadMatrix()
fm_train_real = lm.load_numbers('../data/fm_train_real.dat')
fm_test_real = lm.load_numbers('../data/fm_test_real.dat')
label_train_multiclass = lm.load_labels('../data/label_train_multiclass.dat')
parameter_list = [[
fm_train_real, fm_test_real, label_train_multiclass, 1.2, 1.2, 1e-5, 1,
0.001, 1.5
],
[
fm_train_real, fm_test_real, label_train_multiclass, 5,
1.2, 1e-2, 1, 0.001, 2
]]
def mkl_multiclass_modular(fm_train_real, fm_test_real, label_train_multiclass,
width, C, epsilon, num_threads, mkl_epsilon,
mkl_norm):
from shogun.Features import CombinedFeatures, RealFeatures, Labels
from shogun.Kernel import CombinedKernel, GaussianKernel, LinearKernel, PolyKernel
from shogun.Classifier import MKLMultiClass
kernel = CombinedKernel()
feats_train = CombinedFeatures()
feats_test = CombinedFeatures()
subkfeats_train = RealFeatures(fm_train_real)
subkfeats_test = RealFeatures(fm_test_real)
from tools.load import LoadMatrix
lm=LoadMatrix()
traindat = lm.load_numbers('../data/fm_train_real.dat')
testdat = lm.load_numbers('../data/fm_test_real.dat')
label_traindat = lm.load_labels('../data/label_train_twoclass.dat')
parameter_list = [[traindat,testdat,label_traindat,2.1,1,1e-5],[traindat,testdat,label_traindat,2.2,1,1e-5]]
def classifier_libsvm_modular (fm_train_real=traindat,fm_test_real=testdat,label_train_twoclass=label_traindat,width=2.1,C=1,epsilon=1e-5):
from shogun.Features import RealFeatures, BinaryLabels
from shogun.Kernel import GaussianKernel
from shogun.Classifier import LibSVM
feats_train=RealFeatures(fm_train_real)
feats_test=RealFeatures(fm_test_real)
kernel=GaussianKernel(feats_train, feats_train, width)
labels=BinaryLabels(label_train_twoclass)
svm=LibSVM(C, kernel, labels)
svm.set_epsilon(epsilon)
svm.train()
kernel.init(feats_train, feats_test)
labels = svm.apply().get_labels()
supportvectors = sv_idx=svm.get_support_vectors()
alphas=svm.get_alphas()
predictions = svm.apply()
print predictions.get_labels()
return predictions, svm, predictions.get_labels()
from tools.load import LoadMatrix
from sg import sg
lm = LoadMatrix()
traindna = lm.load_dna('../data/fm_train_dna.dat')
cubedna = lm.load_cubes('../data/fm_train_cube.dat')
parameter_list = [[traindna, cubedna, 3, 0, 'n'],
[traindna, cubedna, 4, 0, 'n']]
def distribution_histogram(fm_train=traindna,
fm_cube=cubedna,
order=3,
gap=0,
reverse='n'):
# sg('new_distribution', 'HISTOGRAM')
sg('add_preproc', 'SORTWORDSTRING')
sg('set_features', 'TRAIN', fm_train, 'DNA')
sg('convert', 'TRAIN', 'STRING', 'CHAR', 'STRING', 'WORD', order,
order - 1, gap, reverse)
sg('attach_preproc', 'TRAIN')
# sg('train_distribution')
# histo=sg('get_histogram')
# num_examples=11
# num_param=sg('get_histogram_num_model_parameters')
# for i in xrange(num_examples):
# for j in xrange(num_param):
# In this example a hidden markov model with 3 states and 6 transitions is trained
# on a string data set. After calling the constructor of the HMM class specifying
# the number of states and transitions the model is trained. Via the Baum-Welch
# algorithm the optimal transition and emission probabilities are estimated. The
# best path, i.e. the path with highest probability given the model can then be
# calculated using get_best_path_state.
from tools.load import LoadMatrix
lm = LoadMatrix()
data = lm.load_cubes('../data/fm_train_cube.dat')
parameter_list = [[data, 1, 64, 1e-5, 2, 0, False, 5],
[data, 3, 6, 1e-1, 1, 0, False, 2]]
def distribution_hmm_modular(fm_cube, N, M, pseudo, order, gap, reverse,
num_examples):
from shogun.Features import StringWordFeatures, StringCharFeatures, CUBE
from shogun.Distribution import HMM, BW_NORMAL
charfeat = StringCharFeatures(CUBE)
charfeat.set_features(fm_cube)
feats = StringWordFeatures(charfeat.get_alphabet())
feats.obtain_from_char(charfeat, order - 1, order, gap, reverse)
hmm = HMM(feats, N, M, pseudo)
hmm.train()
hmm.baum_welch_viterbi_train(BW_NORMAL)
num_examples = feats.get_num_vectors()
num_param = hmm.get_num_model_parameters()
from tools.load import LoadMatrix
from numpy import random
lm = LoadMatrix()
random.seed(17)
ground_truth = lm.load_labels('../data/label_train_multiclass.dat')
predicted = lm.load_labels('../data/label_train_multiclass.dat') * 2
parameter_list = [[ground_truth, predicted]]
def evaluation_multiclassaccuracy_modular(ground_truth, predicted):
from shogun.Features import MulticlassLabels
from shogun.Evaluation import MulticlassAccuracy
ground_truth_labels = MulticlassLabels(ground_truth)
predicted_labels = MulticlassLabels(predicted)
evaluator = MulticlassAccuracy()
accuracy = evaluator.evaluate(predicted_labels, ground_truth_labels)
return accuracy
if __name__ == '__main__':
print('MulticlassAccuracy')
evaluation_multiclassaccuracy_modular(*parameter_list[0])
from tools.load import LoadMatrix
from numpy import ushort
from sg import sg
lm=LoadMatrix()
trainword=ushort(lm.load_numbers('../data/fm_test_word.dat'))
testword=ushort(lm.load_numbers('../data/fm_test_word.dat'))
parameter_list=[[trainword,testword,10,1.4],
[trainword,testword,11,1.5]]
def kernel_linearword (fm_train_word=trainword,fm_test_word=testword,
size_cache=10, scale=1.4):
sg('set_features', 'TRAIN', fm_train_word)
sg('set_features', 'TEST', fm_test_word)
sg('set_kernel', 'LINEAR', 'WORD', size_cache, scale)
km=sg('get_kernel_matrix', 'TRAIN')
km=sg('get_kernel_matrix', 'TEST')
return km
if __name__=='__main__':
print('LinearWord')
kernel_linearword(*parameter_list[0])
###########################################################################
# anova kernel
###########################################################################
from tools.load import LoadMatrix
from numpy import double
lm=LoadMatrix()
traindat = double(lm.load_numbers('../data/fm_train_real.dat'))
testdat = double(lm.load_numbers('../data/fm_test_real.dat'))
parameter_list = [[traindat,testdat,2,10], [traindat,testdat,5,10]]
def kernel_anova_modular (fm_train_real=traindat,fm_test_real=testdat,cardinality=2, size_cache=10):
from shogun.Kernel import ANOVAKernel
from shogun.Features import RealFeatures
feats_train=RealFeatures(fm_train_real)
feats_test=RealFeatures(fm_test_real)
kernel=ANOVAKernel(feats_train, feats_train, cardinality, size_cache)
for i in range(0,feats_train.get_num_vectors()):
for j in range(0,feats_train.get_num_vectors()):
k1 = kernel.compute_rec1(i,j)
k2 = kernel.compute_rec2(i,j)
#if abs(k1-k2) > 1e-10:
# print "|%s|%s|" % (k1, k2)
km_train=kernel.get_kernel_matrix()
kernel.init(feats_train, feats_test)
km_test=kernel.get_kernel_matrix()
return km_train, km_test, kernel
#!/usr/bin/env python
from tools.load import LoadMatrix
lm = LoadMatrix()
traindna = lm.load_dna('../data/fm_train_dna.dat')
testdna = lm.load_dna('../data/fm_test_dna.dat')
parameter_list = [[traindna, testdna, 3, 0, False],
[traindna, testdna, 3, 0, False]]
def distance_canberraword(fm_train_dna=traindna,
fm_test_dna=testdna,
order=3,
gap=0,
reverse=False):
from shogun import StringCharFeatures, StringWordFeatures, DNA
from shogun import SortWordString
from shogun import CanberraWordDistance
charfeat = StringCharFeatures(DNA)
charfeat.set_features(fm_train_dna)
feats_train = StringWordFeatures(charfeat.get_alphabet())
feats_train.obtain_from_char(charfeat, order - 1, order, gap, reverse)
preproc = SortWordString()
preproc.init(feats_train)
feats_train.add_preprocessor(preproc)
feats_train.apply_preprocessor()
charfeat = StringCharFeatures(DNA)
charfeat.set_features(fm_test_dna)
from tools.load import LoadMatrix
import numpy
lm = LoadMatrix()
data = lm.load_numbers("../data/fm_train_real.dat")
parameter_list = [[data]]
def converter_multidimensionalscaling_modular(data):
from shogun.Features import RealFeatures
from shogun.Converter import MultidimensionalScaling
from shogun.Distance import EuclidianDistance
features = RealFeatures(data)
distance_before = EuclidianDistance()
distance_before.init(features, features)
converter = MultidimensionalScaling()
converter.set_target_dim(2)
converter.set_landmark(False)
embedding = converter.apply(features)
distance_after = EuclidianDistance()
distance_after.init(embedding, embedding)
distance_matrix_after = distance_after.get_distance_matrix()
distance_matrix_before = distance_before.get_distance_matrix()
return numpy.linalg.norm(distance_matrix_after - distance_matrix_before) / numpy.linalg.norm(distance_matrix_before)
from tools.load import LoadMatrix
from sg import sg
lm=LoadMatrix()
traindna=lm.load_dna('../data/fm_train_dna.dat')
testdna=lm.load_dna('../data/fm_test_dna.dat')
trainlabel=lm.load_labels('../data/label_train_dna.dat')
parameter_list=[[traindna,testdna,trainlabel,10,3,0,'n',False,'FULL'],
[traindna,testdna,trainlabel,11,4,0,'n',False,'FULL']]
def kernel_salzbergstring (fm_train_dna=traindna,fm_test_dna=testdna,
label_train_dna=trainlabel,size_cache=10,
order=3,gap=0,reverse='n',use_sign=False,
normalization='FULL'):
sg('set_features', 'TRAIN', fm_train_dna, 'DNA')
sg('convert', 'TRAIN', 'STRING', 'CHAR', 'STRING', 'WORD', order, order-1, gap, reverse)
sg('set_features', 'TEST', fm_test_dna, 'DNA')
sg('convert', 'TEST', 'STRING', 'CHAR', 'STRING', 'WORD', order, order-1, gap, reverse)
pseudo_pos=1e-1
pseudo_neg=1e-1
sg('new_plugin_estimator', pseudo_pos, pseudo_neg)
sg('set_labels', 'TRAIN', label_train_dna)
sg('train_estimator')
sg('set_kernel', 'SALZBERG', 'WORD', size_cache)
#sg('set_prior_probs', 0.4, 0.6)
sg('set_prior_probs_from_labels', label_train_dna)
km=sg('get_kernel_matrix', 'TRAIN')
# In this example ROC (Receiver Operator Characteristic) is being computed
# for the pair of ground truth toy labels and random labels.
# ROC curve (as matrix) and auROC (area under ROC) is returned.
from tools.load import LoadMatrix
from numpy import random
lm=LoadMatrix()
ground_truth = lm.load_labels('../data/label_train_twoclass.dat')
random.seed(17)
predicted = random.randn(len(ground_truth))
parameter_list = [[ground_truth,predicted]]
def evaluation_rocevaluation_modular(ground_truth, predicted):
from shogun.Features import BinaryLabels
from shogun.Evaluation import ROCEvaluation
ground_truth_labels = BinaryLabels(ground_truth)
predicted_labels = BinaryLabels(predicted)
evaluator = ROCEvaluation()
evaluator.evaluate(predicted_labels,ground_truth_labels)
return evaluator.get_ROC(), evaluator.get_auROC()
if __name__=='__main__':
print('ROCEvaluation')
evaluation_rocevaluation_modular(*parameter_list[0])
from tools.load import LoadMatrix
lm = LoadMatrix()
traindat = lm.load_numbers('../data/fm_train_real.dat')
testdat = lm.load_numbers('../data/fm_test_real.dat')
parameter_list = [[traindat, testdat], [traindat, testdat]]
def distance_chebyshew_modular(fm_train_real=traindat, fm_test_real=testdat):
from shogun.Features import RealFeatures
from shogun.Distance import ChebyshewMetric
feats_train = RealFeatures(fm_train_real)
feats_test = RealFeatures(fm_test_real)
distance = ChebyshewMetric(feats_train, feats_train)
dm_train = distance.get_distance_matrix()
distance.init(feats_train, feats_test)
dm_test = distance.get_distance_matrix()
return distance, dm_train, dm_test
if __name__ == '__main__':
print 'ChebyshewMetric'
distance_chebyshew_modular(*parameter_list[0])
# This is an example for the initialization of a linear kernel on word (2byte)
# data.
from tools.load import LoadMatrix
from numpy import ushort
lm = LoadMatrix()
traindat = ushort(lm.load_numbers("../data/fm_train_word.dat"))
testdat = ushort(lm.load_numbers("../data/fm_test_word.dat"))
parameter_list = [[traindat, testdat, 1.2], [traindat, testdat, 1.2]]
def kernel_linear_word_modular(fm_train_word=traindat, fm_test_word=testdat, scale=1.2):
from shogun.Kernel import LinearKernel, AvgDiagKernelNormalizer
from shogun.Features import WordFeatures
feats_train = WordFeatures(fm_train_word)
feats_test = WordFeatures(fm_test_word)
kernel = LinearKernel(feats_train, feats_train)
kernel.set_normalizer(AvgDiagKernelNormalizer(scale))
kernel.init(feats_train, feats_train)
km_train = kernel.get_kernel_matrix()
kernel.init(feats_train, feats_test)
km_test = kernel.get_kernel_matrix()
return kernel
#!/usr/bin/env python
from tools.load import LoadMatrix
lm = LoadMatrix()
data = lm.load_cubes("../data/fm_train_cube.dat")
parameter_list = [[data, 1, 64, 1e-5, 2, 0, False, 5], [data, 3, 6, 1e-1, 1, 0, False, 2]]
def distribution_hmm_modular(fm_cube, N, M, pseudo, order, gap, reverse, num_examples):
from shogun.Features import StringWordFeatures, StringCharFeatures, CUBE
from shogun.Distribution import HMM, BW_NORMAL
charfeat = StringCharFeatures(CUBE)
charfeat.set_features(fm_cube)
feats = StringWordFeatures(charfeat.get_alphabet())
feats.obtain_from_char(charfeat, order - 1, order, gap, reverse)
hmm = HMM(feats, N, M, pseudo)
hmm.train()
hmm.baum_welch_viterbi_train(BW_NORMAL)
num_examples = feats.get_num_vectors()
num_param = hmm.get_num_model_parameters()
for i in range(num_examples):
for j in range(num_param):
hmm.get_log_derivative(j, i)
best_path = 0
best_path_state = 0
for i in range(num_examples):
from tools.load import LoadMatrix
lm=LoadMatrix()
data=lm.load_cubes('../data/fm_train_cube.dat')
parameter_list=[[data, 1, 64, 1e-5, 2, 0, False, 5], [data, 3, 6, 1e-1, 1, 0, False, 2]]
def distribution_hmm_modular(fm_cube, N, M, pseudo, order, gap, reverse, num_examples):
from shogun.Features import StringWordFeatures, StringCharFeatures, CUBE
from shogun.Distribution import HMM, BW_NORMAL
charfeat=StringCharFeatures(CUBE)
charfeat.set_features(fm_cube)
feats=StringWordFeatures(charfeat.get_alphabet())
feats.obtain_from_char(charfeat, order-1, order, gap, reverse)
hmm=HMM(feats, N, M, pseudo)
hmm.train()
hmm.baum_welch_viterbi_train(BW_NORMAL)
num_examples=feats.get_num_vectors()
num_param=hmm.get_num_model_parameters()
for i in xrange(num_examples):
for j in xrange(num_param):
hmm.get_log_derivative(j, i)
best_path=0
best_path_state=0
for i in xrange(num_examples):
best_path+=hmm.best_path(i)
for j in xrange(N):
best_path_state+=hmm.get_best_path_state(i, j)
realfeat = RealFeatures(fm_train_real)
feats_train = SparseRealFeatures()
feats_train.obtain_from_simple(realfeat)
realfeat = RealFeatures(fm_test_real)
feats_test = SparseRealFeatures()
feats_test.obtain_from_simple(realfeat)
C = 0.9
epsilon = 1e-5
num_threads = 1
labels = Labels(label_train_twoclass)
svm = SVMOcas(C, feats_train, labels)
svm.set_epsilon(epsilon)
svm.parallel.set_num_threads(num_threads)
svm.set_bias_enabled(False)
svm.train()
svm.set_features(feats_test)
svm.classify().get_labels()
if __name__ == "__main__":
from tools.load import LoadMatrix
lm = LoadMatrix()
fm_train_real = lm.load_numbers("../data/fm_train_real.dat")
fm_test_real = lm.load_numbers("../data/fm_test_real.dat")
label_train_twoclass = lm.load_labels("../data/label_train_twoclass.dat")
svmocas()
# This example shows how to compute the Hamming Word Distance for string features.
from tools.load import LoadMatrix
lm=LoadMatrix()
traindna = lm.load_dna('../data/fm_train_dna.dat')
testdna = lm.load_dna('../data/fm_test_dna.dat')
testdat = lm.load_labels('../data/fm_test_real.dat')
parameter_list = [[traindna,testdna,testdat,4,0,False,False],
[traindna,testdna,testdat,3,0,False,False]]
def distance_hammingword_modular (fm_train_dna=traindna,fm_test_dna=testdna,
fm_test_real=testdat,order=3,gap=0,reverse=False,use_sign=False):
from shogun.Features import StringCharFeatures, StringWordFeatures, DNA
from shogun.Preprocessor import SortWordString
from shogun.Distance import HammingWordDistance
charfeat=StringCharFeatures(DNA)
charfeat.set_features(fm_train_dna)
feats_train=StringWordFeatures(charfeat.get_alphabet())
feats_train.obtain_from_char(charfeat, order-1, order, gap, reverse)
preproc=SortWordString()
preproc.init(feats_train)
feats_train.add_preprocessor(preproc)
feats_train.apply_preprocessor()
charfeat=StringCharFeatures(DNA)
charfeat.set_features(fm_test_dna)
feats_test=StringWordFeatures(charfeat.get_alphabet())
# In this example the Histogram algorithm object computes a histogram over all
# 16bit unsigned integers in the features.
from tools.load import LoadMatrix
from sg import sg
lm = LoadMatrix()
traindna = lm.load_dna("../data/fm_train_dna.dat")
cubedna = lm.load_cubes("../data/fm_train_cube.dat")
parameter_list = [[traindna, cubedna, 3, 0, "n"], [traindna, cubedna, 4, 0, "n"]]
def distribution_histogram(fm_train=traindna, fm_cube=cubedna, order=3, gap=0, reverse="n"):
# sg('new_distribution', 'HISTOGRAM')
sg("add_preproc", "SORTWORDSTRING")
sg("set_features", "TRAIN", fm_train, "DNA")
sg("convert", "TRAIN", "STRING", "CHAR", "STRING", "WORD", order, order - 1, gap, reverse)
sg("attach_preproc", "TRAIN")
# sg('train_distribution')
# histo=sg('get_histogram')
# num_examples=11
# num_param=sg('get_histogram_num_model_parameters')
# for i in xrange(num_examples):
# for j in xrange(num_param):
# sg('get_log_derivative %d %d' % (j, i))
print 'LaRank'
from shogun.Features import RealFeatures, Labels
from shogun.Kernel import GaussianKernel
from shogun.Classifier import LaRank
feats_train=RealFeatures(fm_train_real)
feats_test=RealFeatures(fm_test_real)
width=2.1
kernel=GaussianKernel(feats_train, feats_train, width)
C=1
epsilon=1e-5
labels=Labels(label_train_multiclass)
svm=LaRank(C, kernel, labels)
#svm.set_tau(1e-3)
#svm.set_batch_mode(False)
#svm.io.enable_progress()
svm.set_epsilon(epsilon)
svm.train()
out=svm.classify(feats_train).get_labels()
if __name__=='__main__':
from tools.load import LoadMatrix
lm=LoadMatrix()
fm_train_real=lm.load_numbers('../data/fm_train_real.dat')
fm_test_real=lm.load_numbers('../data/fm_test_real.dat')
label_train_multiclass=lm.load_labels('../data/label_train_multiclass.dat')
larank()
# In this example the distant segments kernel is being computed for toy data.
from tools.load import LoadMatrix
lm = LoadMatrix()
traindat = lm.load_dna("../data/fm_train_dna.dat")
testdat = lm.load_dna("../data/fm_test_dna.dat")
parameter_list = [[traindat, testdat, 5, 5], [traindat, testdat, 6, 6]]
def kernel_distantsegments_modular(fm_train_dna=traindat, fm_test_dna=testdat, delta=5, theta=5):
from shogun.Features import StringCharFeatures, DNA
from shogun.Kernel import DistantSegmentsKernel
feats_train = StringCharFeatures(fm_train_dna, DNA)
feats_test = StringCharFeatures(fm_test_dna, DNA)
kernel = DistantSegmentsKernel(feats_train, feats_train, 10, delta, theta)
km_train = kernel.get_kernel_matrix()
kernel.init(feats_train, feats_test)
km_test = kernel.get_kernel_matrix()
return km_train, km_test, kernel
if __name__ == "__main__":
print("DistantSegments")
kernel_distantsegments_modular(*parameter_list[0])
#
# where \f$\Phi_k\f$ maps a sequence \f${\bf x}\f$ that consists of letters in
# \f$\Sigma\f$ to a feature vector of size \f$|\Sigma|^k\f$. In this feature
# vector each entry denotes how often the k-mer appears in that \f${\bf x}\f$.
#
# Note that this representation is especially tuned to small alphabets
# (like the 2-bit alphabet DNA), for which it enables spectrum kernels
# of order 8.
#
# For this kernel the linadd speedups are quite efficiently implemented using
# direct maps.
#
from tools.load import LoadMatrix
from sg import sg
lm=LoadMatrix()
traindna=lm.load_dna('../data/fm_train_dna.dat')
testdna=lm.load_dna('../data/fm_test_dna.dat')
trainlabel=lm.load_labels('../data/label_train_dna.dat')
parameter_list=[[traindna,testdna,trainlabel,10,3,0,'n',False,'FULL'],
[traindna,testdna,trainlabel,11,4,0,'n',False,'FULL']]
def kernel_weightedcommwordstring (fm_train_dna=traindna,fm_test_dna=testdna,
label_train_dna=trainlabel,size_cache=10,
order=3,gap=0,reverse='n',use_sign=False,
normalization='FULL'):
sg('add_preproc', 'SORTWORDSTRING')
sg('set_features', 'TRAIN', fm_train_dna, 'DNA')
sg('convert', 'TRAIN', 'STRING', 'CHAR', 'STRING', 'WORD', order, order-1, gap, reverse)
epsilon=1e-5
labels=Labels(label_train_twoclass)
svm=LibSVM(C, kernel, labels)
svm.set_epsilon(epsilon)
svm.train()
#kernel.init(feats_train, feats_test)
output = svm.classify(feats_test)#.get_labels()
#output_vector = output.get_labels()
out=svm.classify().get_labels()
testerr=mean(sign(out)!=testlab)
print testerr
#sv_idx=svm.get_support_vectors()
#alphas=svm.get_alphas()
#pm = PerformanceMeasures(output_vector, output)
#acc = pm.get_accuracy()
#roc = pm.get_auROC()
#fms = pm.get_fmeasure()
if __name__=='__main__':
from tools.load import LoadMatrix
lm=LoadMatrix()
fm_train_real=lm.load_numbers('/home/mati/lib/shogun-0.9.3/examples/documented/data/fm_train_real.dat')
fm_test_real=lm.load_numbers('/home/mati/lib/shogun-0.9.3/examples/documented/data/fm_test_real.dat')
label_train_twoclass=lm.load_labels('/home/mati/lib/shogun-0.9.3/examples/documented/data/label_train_twoclass.dat')
libsvm()
#!/usr/bin/env python
from tools.load import LoadMatrix
lm=LoadMatrix()
data=lm.load_numbers('../data/fm_train_real.dat')
label=lm.load_numbers('../data/label_train_twoclass.dat')
parameter_list=[[data,label]]
def features_io_modular (fm_train_real, label_train_twoclass):
import numpy
from modshogun import SparseRealFeatures, RealFeatures, MulticlassLabels
from modshogun import GaussianKernel
from modshogun import LibSVMFile, CSVFile, BinaryFile, HDF5File
feats=SparseRealFeatures(fm_train_real)
feats2=SparseRealFeatures()
f=BinaryFile("tmp/fm_train_sparsereal.bin","w")
feats.save(f)
f=LibSVMFile("tmp/fm_train_sparsereal.ascii","w")
feats.save(f)
f=BinaryFile("tmp/fm_train_sparsereal.bin")
feats2.load(f)
f=LibSVMFile("tmp/fm_train_sparsereal.ascii")
feats2.load(f)
feats=RealFeatures(fm_train_real)
feats2=RealFeatures()
#!/usr/bin/env python
from tools.load import LoadMatrix
from numpy import random
lm=LoadMatrix()
random.seed(17)
ground_truth = lm.load_labels('../data/label_train_multiclass.dat')
predicted = lm.load_labels('../data/label_train_multiclass.dat') * 2
parameter_list = [[ground_truth,predicted]]
def evaluation_multiclassaccuracy_modular (ground_truth, predicted):
from shogun.Features import MulticlassLabels
from shogun.Evaluation import MulticlassAccuracy
ground_truth_labels = MulticlassLabels(ground_truth)
predicted_labels = MulticlassLabels(predicted)
evaluator = MulticlassAccuracy()
accuracy = evaluator.evaluate(predicted_labels,ground_truth_labels)
return accuracy
if __name__=='__main__':
print('MulticlassAccuracy')
evaluation_multiclassaccuracy_modular(*parameter_list[0])
def bray_curtis_distance ():
print 'BrayCurtisDistance'
from sg import sg
sg('set_distance', 'BRAYCURTIS', 'REAL')
sg('set_features', 'TRAIN', fm_train_real)
dm=sg('get_distance_matrix', 'TRAIN')
sg('set_features', 'TEST', fm_test_real)
dm=sg('get_distance_matrix', 'TEST')
if __name__=='__main__':
from tools.load import LoadMatrix
lm=LoadMatrix()
fm_train_real=lm.load_numbers('../data/fm_train_real.dat')
fm_test_real=lm.load_numbers('../data/fm_test_real.dat')
bray_curtis_distance()
def distance():
print "Distance"
width = 1.7
size_cache = 10
from sg import sg
sg("set_features", "TRAIN", fm_train_real)
sg("set_features", "TEST", fm_test_real)
sg("set_distance", "EUCLIDIAN", "REAL")
sg("set_kernel", "DISTANCE", size_cache, width)
km = sg("get_kernel_matrix", "TRAIN")
km = sg("get_kernel_matrix", "TEST")
if __name__ == "__main__":
from tools.load import LoadMatrix
lm = LoadMatrix()
fm_train_real = lm.load_numbers("../data/fm_train_real.dat")
fm_test_real = lm.load_numbers("../data/fm_test_real.dat")
distance()
# In this example a two-class linear support vector machine classifier is trained
# on a toy data set and the trained classifier is used to predict labels of test
# examples. As training algorithm the Stochastic Gradient Descent (SGD) solver is
# used with the SVM regularization parameter C=0.9. The number of iterations, i.e.
# passes though all training examples, is set to num_iter=5 .
#
# For more details on the SGD solver see
# L. Bottou, O. Bousquet. The tradeoff of large scale learning. In NIPS 20. MIT
# Press. 2008.
from tools.load import LoadMatrix
lm=LoadMatrix()
traindat = lm.load_numbers('../data/fm_train_real.dat')
testdat = lm.load_numbers('../data/fm_test_real.dat')
label_traindat = lm.load_labels('../data/label_train_twoclass.dat')
parameter_list = [[traindat,testdat,label_traindat,0.9,1,6],[traindat,testdat,label_traindat,0.8,1,5]]
def classifier_svmsgd_modular (fm_train_real=traindat,fm_test_real=testdat,label_train_twoclass=label_traindat,C=0.9,num_threads=1,num_iter=5):
from shogun.Features import RealFeatures, SparseRealFeatures, Labels
from shogun.Classifier import SVMSGD
realfeat=RealFeatures(fm_train_real)
feats_train=SparseRealFeatures()
feats_train.obtain_from_simple(realfeat)
realfeat=RealFeatures(fm_test_real)
feats_test=SparseRealFeatures()
feats_test.obtain_from_simple(realfeat)
from tools.load import LoadMatrix
from sg import sg
lm=LoadMatrix()
traindat=lm.load_numbers('../data/fm_train_real.dat')
testdat=lm.load_numbers('../data/fm_test_real.dat')
train_label=lm.load_labels('../data/label_train_multiclass.dat')
parameter_list=[[traindat,testdat, train_label,10,2.1,1.2,1e-5,False],
[traindat,testdat,train_label,10,2.1,1.3,1e-4,False]]
def classifier_gmnpsvm (fm_train_real=traindat,fm_test_real=testdat,
label_train_multiclass=train_label,
size_cache=10, width=2.1,C=1.2,
epsilon=1e-5,use_bias=False):
sg('set_features', 'TRAIN', fm_train_real)
sg('set_kernel', 'GAUSSIAN', 'REAL', size_cache, width)
sg('set_labels', 'TRAIN', label_train_multiclass)
sg('new_classifier', 'GMNPSVM')
sg('svm_epsilon', epsilon)
sg('c', C)
sg('svm_use_bias', use_bias)
sg('train_classifier')
sg('set_features', 'TEST', fm_test_real)
result=sg('classify')
kernel_matrix = sg('get_kernel_matrix', 'TEST')
return result, kernel_matrix
from tools.load import LoadMatrix
lm=LoadMatrix()
data = lm.load_numbers('../data/fm_train_real.dat')
parameter_list = [[data, 0.01, 1.0], [data, 0.05, 2.0]]
def preprocessor_kernelpcacut_modular(data, threshold, width):
from shogun.Features import RealFeatures
from shogun.Preprocessor import KernelPCACut
from shogun.Kernel import GaussianKernel
features = RealFeatures(data)
kernel = GaussianKernel(features,features,width)
preprocessor = KernelPCACut(kernel,threshold)
preprocessor.init(features)
preprocessor.apply_to_feature_matrix(features)
return features
if __name__=='__main__':
print 'KernelPCACut'
preprocessor_kernelpcacut_modular(*parameter_list[0])
#!/usr/bin/env python
from tools.load import LoadMatrix
lm = LoadMatrix()
train_dna = lm.load_dna("../data/fm_train_dna.dat")
test_dna = lm.load_dna("../data/fm_test_dna.dat")
label = lm.load_labels("../data/label_train_dna.dat")
parameter_list = [[train_dna, test_dna, label, 20, 0.9, 1e-3, 1], [train_dna, test_dna, label, 20, 2.3, 1e-5, 4]]
def classifier_svmlight_batch_linadd_modular(
fm_train_dna, fm_test_dna, label_train_dna, degree, C, epsilon, num_threads
):
from modshogun import StringCharFeatures, BinaryLabels, DNA
from modshogun import WeightedDegreeStringKernel, MSG_DEBUG
try:
from modshogun import SVMLight
except ImportError:
print("No support for SVMLight available.")
return
feats_train = StringCharFeatures(DNA)
# feats_train.io.set_loglevel(MSG_DEBUG)
feats_train.set_features(fm_train_dna)
feats_test = StringCharFeatures(DNA)
feats_test.set_features(fm_test_dna)
degree = 20
# This is an example for the initialization of the CommWordString-kernel (aka
# Spectrum or n-gram kernel; its name is derived from the unix command comm). This kernel
# sums over k-mere matches (k='order'). For efficient computing a preprocessor is used
# that extracts and sorts all k-mers. If 'use_sign' is set to one each k-mere is counted
# only once.
from tools.load import LoadMatrix
from sg import sg
lm=LoadMatrix()
traindna=lm.load_dna('../data/fm_train_dna.dat')
testdna=lm.load_dna('../data/fm_test_dna.dat')
parameter_list=[[traindna,testdna,10,3,0,'n',False,'FULL'],
[traindna,testdna,11,4,0,'n',False,'FULL']]
def kernel_commwordstring (fm_train_dna=traindna,fm_test_dna=testdna,
size_cache=10,
order=3,gap=0,reverse='n',
use_sign=False,normalization='FULL'):
sg('add_preproc', 'SORTWORDSTRING')
sg('set_features', 'TRAIN', fm_train_dna, 'DNA')
sg('convert', 'TRAIN', 'STRING', 'CHAR', 'STRING', 'WORD', order, order-1, gap, reverse)
sg('attach_preproc', 'TRAIN')
sg('set_features', 'TEST', fm_test_dna, 'DNA')
sg('convert', 'TEST', 'STRING', 'CHAR', 'STRING', 'WORD', order, order-1, gap, reverse)
sg('attach_preproc', 'TEST')
sg('set_kernel', 'COMMSTRING', 'WORD', size_cache, use_sign, normalization)
km=sg('get_kernel_matrix', 'TRAIN')
# This is an example for the initialization of a linear kernel on raw byte
# data.
###########################################################################
# linear kernel on byte features
###########################################################################
from tools.load import LoadMatrix
from numpy import ubyte
lm = LoadMatrix()
traindat = ubyte(lm.load_numbers('../data/fm_train_byte.dat'))
testdat = ubyte(lm.load_numbers('../data/fm_test_byte.dat'))
parameter_list = [[traindat, testdat], [traindat, testdat]]
def kernel_linear_byte_modular(fm_train_byte=traindat, fm_test_byte=testdat):
from shogun.Kernel import LinearKernel
from shogun.Features import ByteFeatures
feats_train = ByteFeatures(fm_train_byte)
feats_test = ByteFeatures(fm_test_byte)
kernel = LinearKernel(feats_train, feats_train)
km_train = kernel.get_kernel_matrix()
kernel.init(feats_train, feats_test)
km_test = kernel.get_kernel_matrix()
return kernel
