def main():
"""
In pseudo-code:
load data
create graphs from data
for {kfoldcross times}
do {take sample from training data}
predict results on remainder (= proteins not in sample)
measure results on remainder
take average f-measure
Parameters
----------
See args.
"""
data = load_data()
hgraph = nx.Graph(data.humanppi)
fgraph = nx.Graph(data.functions)
if args.test2:
print "run predictions on test2 set"
proteins_test2 = data.test2
predictions = predict_cps(proteins_test2, hgraph, fgraph)
print zip(proteins_test2, predictions)
write_test2_prediction(proteins_test2, predictions)
else:
sampler = Sampler(data.test1, args.samplesize)
results = list() #to be populated by Results class objects
for i in range(args.kfoldcross):
sample = zip(*sampler.remainder)[0]
predictions = predict_cps(sample, hgraph, fgraph)
result = Results(sampler.remainder, predictions)
results.append(result)
sampler.resample()
if args.verbose:
result.print_results()
result.print_confusion_matrix()
f_measures = [i.f_measure for i in results]
avg_f_measure = sum(f_measures)/float(len(f_measures))
print "\n\nAverage F-Measure:", avg_f_measure
pause = raw_input("Hit Enter to continue... ")
return 0
def sample(trained_weights_path, trained_model_name, trained_weights_name,
input_path, output_path, search_method):
meta_dataset = np.load("{}/meta_dataset.npy".format(trained_weights_path))
input_shape = meta_dataset[0]
output_size = meta_dataset[1]
vocabulary_size = meta_dataset[2]
# voc = Vocabulary()
# voc.retrieve(trained_weights_path)
# vocabulary_size = voc.size
# model = pix2code(input_shape, output_size, trained_weights_path)
model = newpix2code(input_shape, output_size, trained_weights_path,
vocabulary_size)
# trained_weights_name 为带后缀名的，如 .hdf5
model.load(name=trained_model_name, weights_name=trained_weights_name)
sampler = Sampler(trained_weights_path, input_shape, output_size,
CONTEXT_LENGTH)
file_name = basename(input_path)[:basename(input_path).find(".")]
evaluation_img = Utils.get_preprocessed_img(input_path, IMAGE_SIZE)
if search_method == "greedy":
result, _ = sampler.predict_greedy(model,
np.array([evaluation_img]),
require_sparse_label=False)
print("Result greedy: {}".format(result))
else:
beam_width = int(search_method)
print("Search with beam width: {}".format(beam_width))
result, _ = sampler.predict_beam_search(model,
np.array([evaluation_img]),
beam_width=beam_width,
require_sparse_label=False)
print("Result beam: {}".format(result))
with open("{}/{}.gui".format(output_path, file_name), 'w') as out_f:
out_f.write(result.replace(START_TOKEN, "").replace(END_TOKEN, ""))
trained_weights_path = argv[0]
trained_model_name = argv[1]
input_path = argv[2]
output_path = argv[3]
encoding_type = argv[4]
search_method = "greedy" if len(argv) < 6 else argv[5]
meta_dataset = np.load("{}/meta_dataset.npy".format(trained_weights_path),
allow_pickle=True)
input_shape = meta_dataset[0]
output_size = meta_dataset[1]
model = pix2code(input_shape, output_size, trained_weights_path, encoding_type)
model.load(trained_model_name)
sampler = Sampler(trained_weights_path, input_shape, output_size,
CONTEXT_LENGTH)
# dataset = Dataset()
# if encoding_type == "one_hot":
# dataset.load_with_one_hot_encoding(input_path, generate_binary_sequences=True)
# elif encoding_type == "w2v":
# dataset.load_with_word2vec(input_path, generate_binary_sequences=True)
# else:
# raise Exception("Missing parameter")
# dataset = Dataset()
# dataset.load(input_path))
''' Metodo di valutazione cinese
model.compile()
score, loss = model.minevaluate(dataset)
print("accuracy: ", score)
np_load_old = np.load
# modify the default parameters of np.load
np.load = lambda *a,**k: np_load_old(*a, allow_pickle=True, **k)
# restore np.load for future normal usage
np.load = np_load_old
meta_dataset = np.load("{}/meta_dataset.npy".format(trained_weights_path))
input_shape = meta_dataset[0]
output_size = meta_dataset[1]
model = pix2code(input_shape, output_size, trained_weights_path)
model.load(trained_model_name)
sampler = Sampler(trained_weights_path, input_shape, output_size, CONTEXT_LENGTH)
file_name = basename(input_path_tablet)[:basename(input_path_tablet).find(".")]
#adjusted in order to deal with two input images
evaluation_img_tablet = Utils.get_preprocessed_img(input_path_tablet, IMAGE_SIZE)
evaluation_img_desktop = Utils.get_preprocessed_img(input_path_desktop, IMAGE_SIZE)
if search_method == "greedy":
result, _ = sampler.predict_greedy(model, np.array([evaluation_img_tablet]), np.array([evaluation_img_desktop]))
print("Result greedy: {}".format(result))
else:
beam_width = int(search_method)
print("Search with beam width: {}".format(beam_width))
result, _ = sampler.predict_beam_search(model, np.array([evaluation_img_tablet]), np.array([evaluation_img_desktop]), beam_width=beam_width)
print("Result beam: {}".format(result))
def evaluate_model(trained_weights_path, trained_model_name, trained_weights_name, input_path, output_path,search_method="greedy"):
meta_dataset = np.load("{}/meta_dataset.npy".format(trained_weights_path))
input_shape = meta_dataset[0]
output_size = meta_dataset[1]
# get vocabulary size
voc = Vocabulary()
voc.retrieve(weights_path)
vocabulary_size = voc.size
model = UItoPage(input_shape, output_size, trained_weights_path, vocabulary_size)
# model.load(trained_model_name)
# trained_weights_name 为带后缀名的，如 xx.hdf5
model.load(name=trained_model_name, weights_name=trained_weights_name)
sampler = Sampler(trained_weights_path, input_shape, output_size, CONTEXT_LENGTH)
actual, predicted = list(), list()
current_num = 0
for f in os.listdir(input_path):
if f.find(".gui") != -1:
file_name = f[:f.find(".gui")]
evaluation_img = None
if os.path.isfile("{}/{}.png".format(input_path, file_name)):
evaluation_img = Utils.get_preprocessed_img("{}/{}.png".format(input_path, file_name), IMAGE_SIZE)
# 若output_path是提前predict过的所有测试集的结果，则使用已经已有的.gui文件
if os.path.isfile("{}/{}.gui".format(output_path, file_name)):
predict_result = load_doc("{}/{}.gui".format(output_path, file_name))
else:
if evaluation_img is not None:
if search_method == "greedy":
result, _ = sampler.predict_greedy(model, np.array([evaluation_img]),require_sparse_label=False)
predict_result = result.replace(START_TOKEN, "").replace(END_TOKEN, "")
current_num += 1
print("Num:{} Evaluate image: {}\n".format(current_num,file_name))
else:
beam_width = int(search_method)
print("Search with beam width: {}".format(beam_width))
result, _ = sampler.predict_beam_search(model, np.array([evaluation_img]), beam_width=beam_width, require_sparse_label=False)
predict_result =result.replace(START_TOKEN, "").replace(END_TOKEN, "")
current_num += 1
print("Num:{} Evaluate image: {}\n".format(current_num,file_name))
with open("{}/{}.gui".format(output_path, file_name), 'w') as out_f:
out_f.write(predict_result)
else:
print("No image exists in input path.")
continue
predict_result = ' '.join(predict_result.split())
predict_result = predict_result.replace(",", " , ").replace("}", " } ").replace("{", " { ")
predict_result = predict_result.replace(" ", " ").replace(" ", " ")
predicted.append(predict_result.split())
actual_gui = load_doc("{}/{}.gui".format(input_path, file_name))
actual_gui = ' '.join(actual_gui.split())
actual_gui = actual_gui.replace(",", " , ").replace("}", " } ").replace("{", " { ")
actual_gui = actual_gui.replace(" ", " ").replace(" ", " ")
actual.append([actual_gui.split()])
assert len(actual) == len(predicted)
bleu = corpus_bleu(actual, predicted)
return bleu, actual, predicted
np_load_old = np.load
# modify the default parameters of np.load
np.load = lambda *a, **k: np_load_old(*a, allow_pickle=True, **k)
# restore np.load for future normal usage
np.load = np_load_old
meta_dataset = np.load("{}/meta_dataset.npy".format(trained_weights_path))
input_shape = meta_dataset[0]
output_size = meta_dataset[1]
model = pix2code(input_shape, output_size, trained_weights_path)
model.load(trained_model_name)
sampler = Sampler(trained_weights_path, input_shape, output_size,
CONTEXT_LENGTH)
file_name = basename(input_path_tablet)[:basename(input_path_tablet).find(".")]
#adjusted in order to deal with two input images
evaluation_img_tablet = Utils.get_preprocessed_img(input_path_tablet,
IMAGE_SIZE)
evaluation_img_desktop = Utils.get_preprocessed_img(input_path_desktop,
IMAGE_SIZE)
result = sampler.complete_sequence(model, np.array([evaluation_img_tablet]),
np.array([evaluation_img_desktop]),
input_path_gui)
print("Result greedy: {}".format(result))
with open("{}/{}.gui".format(output_path, file_name), 'w') as out_f:
out_f.write(result.replace(START_TOKEN, "").replace(END_TOKEN, ""))