def accuracy(self, examples, descriptions):
dst, ind = self.knn.query(examples, self.k)
consensus_captions = [] # consensus captions
actual_descriptions = [] # actual descriptions for all examples
print('computing accuracy')
for i, ids in enumerate(ind):
captions = [] # candidate captions
for j, id in enumerate(ids):
# get consensus captions
sentences = self.trains['descriptions'][id]['sentences']
for sentence in sentences:
captions.append(sentence['raw'])
consensus_captions.append(consensus(captions))
sentences = descriptions[i]["sentences"]
actual_description = []
for sentence in sentences:
actual_description.append(sentence['raw'])
actual_descriptions.append(actual_description)
if i % 10 == 0:
print("{0:.0f}%".format(float(i) / len(examples) * 100))
print('100%')
print(accuracy(consensus_captions, actual_descriptions))
def eval(self, examples, descriptions):
'''
test knn model and display image
:return:
'''
dst, ind = self.knn.query(examples, self.k)
consensus_captions = [] # consensus captions
actual_descriptions = [] # actual descriptions for all examples
for i, ids in enumerate(ind):
print('{0} predict image {1} filename: {2} {0}'.format(
banner1, i, descriptions[i]['filename']))
captions = [] # candidate captions
for j, id in enumerate(ids):
print('{0} closest image: {1} image id: {2} {0}'.format(
banner2, j, self.trains['descriptions'][id]['filename']))
sentences = self.trains['descriptions'][id]['sentences']
for sentence in sentences:
print sentence['raw']
#get consensus captions
sentences = self.trains['descriptions'][id]['sentences']
for sentence in sentences:
captions.append(sentence['raw'])
print("Predictions: ", consensus(captions))
consensus_captions.append(consensus(captions))
sentences = descriptions[i]["sentences"]
print('actual description:')
actual_description = []
for sentence in sentences:
actual_description.append(sentence['raw'])
print(sentence['raw'])
actual_descriptions.append(actual_description)
img = mpimg.imread('data/Flicker8k_Dataset/' +
descriptions[i]['filename'])
plt.imshow(img)
plt.show()
return consensus_captions, actual_descriptions
def predict(self, examples):
dst, ind = self.knn.query(examples, self.k)
consensus_captions = [] #consensus captions
for i, ids in enumerate(ind):
print('predict image ', i)
captions = [] # candidate captions
for j, id in enumerate(ids):
# print 'closed image:', j, 'image id:', id
sentences = self.trains['descriptions'][id]['sentences']
for sentence in sentences:
captions.append(sentence['raw'])
consensus_captions.append(consensus(captions))
return consensus_captions
def predict(self, examples):
'''
get the k nearest neighbors of given examples
:param examples: images needed to be described
:return:
'''
dst, ind = self.knn.query(examples, self.k)
consensus_captions = [] #consensus captions
for i, ids in enumerate(ind):
print('predict image ', i)
captions = [] # candidate captions
for j, id in enumerate(ids):
# print 'closed image:', j, 'image id:', id
sentences = self.trains['descriptions'][id]['sentences']
for sentence in sentences:
captions.append(sentence['raw'])
consensus_captions.append(consensus(captions))
return consensus_captions
samples = samples[minflag]
mutations_temp = numpy.zeros((len(mutations),len(genes)))
mutations_temp[:,tokeep_geneindex] = mutations[:,tokeep_thisindex]
mutations = sp.csr_matrix(mutations_temp)
mutation_smooth = utils.diffuse(mutations,data['adj'],alpha,diff_thresh)
mutation_smooth_norm = sp.csr_matrix(utils.quantile_normalization(numpy.array(mutation_smooth.todense())),shape=mutation_smooth.shape)
#U,V = utils.gnmf(mutation_smooth,data['knn'],nclust, gamma, maxiter, tolerance)
#labels = numpy.array(V.todense().argmax(axis=1))[:,0]
def gnmfsingle(X, W, nclust, gamma, maxiter, tolerance):
U,V = utils.gnmf(X, W ,nclust, gamma, maxiter, tolerance)
return numpy.array(V.todense().argmax(axis=1))[:,0]
cons = utils.consensus(gnmfsingle,mutation_smooth_norm, [data['knn'],nclust, gamma, maxiter, tolerance], bootstrap = 0.8,rep = 100)
######take from stratipy modules
zmatrix = linkage(cons,method='average')
clusters = fcluster(zmatrix,1)
dend = dendrogram(zmatrix,count_sort='ascending')
idx=numpy.array(dend['leaves'])
plt.imshow(cons[idx,:][:,idx])
plt.plot(range(1, len(zmatrix)+1), zmatrix[::-1, 2])
knee = numpy.diff(zmatrix[::-1, 2], 2)
plt.plot(range(2, len(zmatrix)), knee)
num_clust1 = knee.argmax() + 2
knee[knee.argmax()] = 0
return individual
if __name__ == "__main__":
instance = 'x60189_4'
matrix = np.genfromtxt(instance + '/matrix_conservative.csv',
delimiter=',')
#print(matrix.shape)
num_fragments = matrix.shape[0]
aleatory_solution = np.arange(num_fragments)
np.random.shuffle(aleatory_solution)
print("initial solution: ", aleatory_solution)
sol = PALS(num_fragments, aleatory_solution, matrix)
fitness_temp = fitness(matrix, sol)
contigs_temp = consensus(matrix, sol)
print("fitness: ", fitness_temp, "contig: ", contigs_temp)
print("final solution: ", sol)
#################################################### pruebas #########################################
num_test = 30.0
fitness_acum = best_fitness = contig_acum = 0.0
best_contig = 100
"""
print("TESTING 30 ITEARTIONS...")
for i in range(int(num_test)):
aleatory_solution = np.arange(num_fragments)
np.random.shuffle(aleatory_solution)
print("testing...", i)