def kernel_weighted_degree_string_modular (fm_train_dna=traindat,fm_test_dna=testdat,degree=20):
from shogun.Features import StringCharFeatures, DNA
from shogun.Kernel import WeightedDegreeStringKernel, MSG_DEBUG
feats_train=StringCharFeatures(fm_train_dna, DNA)
#feats_train.io.set_loglevel(MSG_DEBUG)
feats_test=StringCharFeatures(fm_test_dna, DNA)
kernel=WeightedDegreeStringKernel(feats_train, feats_train, degree)
from numpy import arange,double
weights=arange(1,degree+1,dtype=double)[::-1]/ \
sum(arange(1,degree+1,dtype=double))
kernel.set_wd_weights(weights)
#from numpy import ones,float64,int32
#kernel.set_position_weights(ones(len(fm_train_dna[0]), dtype=float64))
km_train=kernel.get_kernel_matrix()
kernel.init(feats_train, feats_test)
km_test=kernel.get_kernel_matrix()
#this is how to serializate the kernel
#import pickle
#pickle.dump(kernel, file('kernel_obj.dump','w'), protocol=2)
#k=pickle.load(file('kernel_obj.dump','r'))
return km_train, km_test, kernel
def weighted_degree_string (): print 'WeightedDegreeString' from shogun.Features import StringCharFeatures, DNA from shogun.Kernel import WeightedDegreeStringKernel feats_train=StringCharFeatures(fm_train_dna, DNA) feats_test=StringCharFeatures(fm_test_dna, DNA) degree=20 kernel=WeightedDegreeStringKernel(feats_train, feats_train, degree) #weights=arange(1,degree+1,dtype=double)[::-1]/ \ # sum(arange(1,degree+1,dtype=double)) #kernel.set_wd_weights(weights) km_train=kernel.get_kernel_matrix() kernel.init(feats_train, feats_test) km_test=kernel.get_kernel_matrix()
normalizer = MultitaskKernelNormalizer(task_vector)
#wdk.set_task_vector(task_vector) #, task_vector)
for i in xrange(2):
for j in xrange(2):
if i==j:
normalizer.set_task_similarity(i,j, 4.0)
else:
normalizer.set_task_similarity(i,j, 1.0)
base_wdk.set_normalizer(normalizer)
print base_wdk.get_kernel_matrix()
print "--->",base_wdk.get_normalizer().get_name()
wdk = WeightedDegreeStringKernel(feat, feat, 1)
normalizer = IdentityKernelNormalizer()
wdk.set_normalizer(normalizer)
##################################################################
# external modification
##################################################################
km = wdk.get_kernel_matrix()
for i in xrange(N):
normalizer = MultitaskKernelNormalizer(task_vector)
#wdk.set_task_vector(task_vector) #, task_vector)
for i in xrange(2):
for j in xrange(2):
if i == j:
normalizer.set_task_similarity(i, j, 4.0)
else:
normalizer.set_task_similarity(i, j, 1.0)
base_wdk.set_normalizer(normalizer)
print base_wdk.get_kernel_matrix()
print "--->", base_wdk.get_normalizer().get_name()
wdk = WeightedDegreeStringKernel(feat, feat, 1)
normalizer = IdentityKernelNormalizer()
wdk.set_normalizer(normalizer)
##################################################################
# external modification
##################################################################
km = wdk.get_kernel_matrix()
for i in xrange(N):
for j in xrange(N):
for j in xrange(N):
external_objective += alphas_full[j] * f[j]
for k in xrange(N):
external_objective += 0.5 * alphas_full[j] * alphas_full[k] * y[j] * y[k] * wdk.kernel(j,k)
print "svmlight external objective:", external_objective
#precompute kernel matrix using shogun
K = wdk.get_kernel_matrix()
K = numpy.transpose(y.flatten() * (K*y.flatten()).transpose())
C = 1.0
# Important!! QP does not accept ndarray as a type, it must be an array
p = QP(K, f, Aeq=y, beq=0, lb=numpy.zeros(N), ub=C*numpy.ones(N))
r = p.solve('cvxopt_qp', iprint = 0)
print "cvxopt objective:", r.ff
print "cvxopt alphas:", numpy.array(r.xf[0:5])
external_objective = 0.0
for j in xrange(N):