def main():
model_dir = './log/model'
device_name = "/cpu:0"
log_dir = './log/'
data_dir = 'MNIST_data'
data_set = input_data.read_data_sets(data_dir, one_hot=True)
# build graph
with tf.Graph().as_default() as g:
with tf.device(device_name):
(x, y_), loss, accuracy, saver = \
build_graph(is_learning=False, enable_bn=False)
# run session
with tf.Session(graph=g) as sess:
# restore model
saver.restore(sess, model_dir)
writer = tf.summary.FileWriter(log_dir)
writer.add_graph(sess.graph)
merged_summary = tf.summary.merge(
[tf.summary.merge_all('all'),
tf.summary.merge_all('layers')])
s = sess.run(fetches=merged_summary,
feed_dict={
x: [data_set.test.images[0]],
y_: [data_set.test.labels[0]]
})
# write log
writer.add_summary(s, 0)
def detect_bridge_cells(year):
G = build_graph(year)
# read annual data
with open('../resources/' + year + '_cell_data.json', 'r') as f:
cells = json.load(f)
border_cells = {k for k in cells if cells[k]['border']}
live_cells = {k for k in cells if cells[k]['live']}
all_cells = border_cells.union(live_cells)
bridge_cells = set()
for cell in live_cells:
neighborhood = cells_within([cell], cell, set(), RADIUS)
for neighbor in neighborhood:
if neighbor in live_cells:
path = networkx.shortest_path(G, cell, neighbor)[1:-1]
for path_cell in path:
if path_cell in border_cells:
bridge_cells.add(path_cell)
for cell in bridge_cells:
cells[cell]['bridge'] = True
# write annual data
with open('../resources/' + year + '_cell_data.json', 'w') as f:
f.write(
json.dumps(cells, indent=4, sort_keys=True, separators=(',', ':')))
def main():
parser = argparse.ArgumentParser(description="graph_stats provides some basic statistics about directed graphs")
parser.add_argument(dest="fname", help="the name of the graph file")
parser.add_argument(
"-n",
action="store",
dest="num_display",
default=10,
type=int,
help="Number of users to display in the statistics",
)
parser.add_argument(
"--prune", action="store_true", default=False, help="Do NOT include leaf nodes in the graph calculations"
)
parser.add_argument("--verbose", action="store_true", default=False, help="Enable verbose output")
results = parser.parse_args()
graph = build_graph.build_graph(results.fname, results.prune, results.verbose)
print "Graph Statistics:"
print "Total Nodes " + str(graph.get_num_users())
print "Total Edges " + str(graph.get_num_edges())
graph.sort_by_links()
for i in xrange(0, min(results.num_display, graph.get_num_users())):
most_linked = graph.get_user_linked(i)
print str(i + 1) + "th most linked " + most_linked.name + " with " + str(len(most_linked.followers))
graph.calculate_friends()
for i in xrange(0, min(results.num_display, graph.get_num_users())):
most_friends = graph.get_user_friends(i)
print str(i + 1) + "th most friends " + most_friends.name + " with " + str(len(most_friends.friends))
print "Average Friends " + str(graph.avg_friends)
def build(constants):
if constants.ACTIVATION == 'leaky_relu':
activation = tf.nn.leaky_relu
elif constants.ACTIVATION == 'relu':
activation = tf.nn.relu
elif constants.ACTIVATION == 'tanh':
activation = tf.nn.tanh
else:
activation = tf.nn.relu
# make networks
encoder = make_encoder(constants.CONVS, constants.FCS, activation)
decoder = make_decoder(constants.CONVS, constants.FCS, activation)
sample_latent = make_latent(constants.LATENT_SIZE)
# build graphs
reconstruct,\
generate,\
train = build_graph(
encoder=encoder,
decoder=decoder,
sample_latent=sample_latent,
image_size=constants.IMAGE_SIZE,
latent_size=constants.LATENT_SIZE,
lr=constants.LR
)
return reconstruct, generate, train
def main():
filename = sys.argv[1]
num_runs = int(sys.argv[2])
G = build_graph.build_graph(filename)
out_prefix = filename.split(".")[0]
mccabe = open(out_prefix + "_mccabe", "a")
algo2 = open(out_prefix + "_algo2", "a")
algo3 = open(out_prefix + "_algo3", "a")
paper = open(out_prefix + "_paper", "a")
total_mccabe = 0
total_algo2 = 0
total_algo3 = 0
total_paper = 0
for i in range(num_runs):
total_mccabe += time_mccabe(G, mccabe)
total_algo2 += time_algo2(G, algo2)
total_algo3 += time_algo3(G, algo3)
total_paper += time_paper_algo(G, paper)
mccabe.write(str(total_mccabe / num_runs))
algo2.write(str(total_algo2 / num_runs))
algo3.write(str(total_algo3 / num_runs))
paper.write(str(total_paper / num_runs))
mccabe.close()
algo2.close()
algo3.close()
paper.close()
def main():
# get data
# (graph 1)
# build graph
# start training
# init global variables
# merge all summaries
# iterate training set
# if i % summary_period == 0 or i+1 == max_epochs
# run summary tensor
#
# run train tensor
#
# save model
#
# (graph 2)
# build graph
# start testing
# restore model
# eval accuracy of testing set
data_set = input_data.read_data_sets(FLAGS.data_dir, one_hot=True)
with tf.Graph().as_default() as g_train:
with tf.device(FLAGS.device_name):
(x, y_), loss, accuracy, saver = \
build_graph(is_learning=True, enable_bn=True)
with tf.Session(graph=g_train) as sess:
train(x, y_, loss, FLAGS.lr, accuracy,
data_set, FLAGS.batch_size, FLAGS.max_steps,
FLAGS.summary_period, FLAGS.print_period,
sess, saver, FLAGS.model_dir, FLAGS.training_log_dir)
with tf.Graph().as_default() as g_test:
with tf.device(FLAGS.device_name):
(x, y_), loss, accuracy, saver = \
build_graph(is_learning=False, enable_bn=True)
with tf.Session(graph=g_test) as sess:
test(x, y_, accuracy, data_set,
saver, FLAGS.model_dir, sess)
def get_data_from_provider_update_db(connection, control):
# Получаем данные от провайдера
p = Provider()
data = p.get_data()
data = build_graph(data)
# Записываем данные в базу
cur = connection.cursor()
r = control.upsert_request(data)
cur.execute(r)
conn.commit()
cur.close()
return data
def __init__(self,
actor,
critic,
value,
obs_dim,
num_actions,
replay_buffer,
batch_size=4,
action_scale=2.0,
gamma=0.9,
tau=0.01,
actor_lr=3 * 1e-3,
critic_lr=3 * 1e-3,
value_lr=3 * 1e-3,
reg_factor=1e-3):
self.batch_size = batch_size
self.num_actions = num_actions
self.gamma = gamma
self.obs_dim = obs_dim
self.action_scale = action_scale
self.last_obs = None
self.t = 0
self.replay_buffer = replay_buffer
self._act,\
self._train_actor,\
self._train_critic,\
self._train_value,\
self._update_target = build_graph(
actor=actor,
critic=critic,
value=value,
obs_dim=obs_dim,
num_actions=num_actions,
batch_size=batch_size,
gamma=gamma,
tau=tau,
actor_lr=actor_lr,
critic_lr=critic_lr,
value_lr=value_lr,
reg_factor=reg_factor
)
self.actor_errors = []
self.critic_errors = []
self.value_errors = []
def main(path, mode):
adjacency, offsets, test_df, user2uid, item2iid = build_graph(
path, test_size=32, split="time", out_dir="model"
)
n_users, n_items = len(user2uid), len(item2iid)
latest_prefs = test_df.groupby("user_id").item_id.apply(list).to_dict()
rw_class = {
"vanilla": RW,
"cache": RWcache,
"parallel": RWparallel,
"cy": RWcy,
"cython": RWcython,
}
rw = rw_class[mode](alpha=0.1, n_total_steps=100000).load_graph(
adjacency, offsets, n_users, n_items
)
run_evaluate(rw, latest_prefs, 32)
def main():
parser = argparse.ArgumentParser(
description=
'graph_stats provides some basic statistics about directed graphs')
parser.add_argument(dest="fname", help="the name of the graph file")
parser.add_argument('-n',
action="store",
dest="num_display",
default=10,
type=int,
help="Number of users to display in the statistics")
parser.add_argument(
'--prune',
action="store_true",
default=False,
help='Do NOT include leaf nodes in the graph calculations')
parser.add_argument('--verbose',
action="store_true",
default=False,
help='Enable verbose output')
results = parser.parse_args()
graph = build_graph.build_graph(results.fname, results.prune,
results.verbose)
print "Graph Statistics:"
print "Total Nodes " + str(graph.get_num_users())
print "Total Edges " + str(graph.get_num_edges())
graph.sort_by_links()
for i in xrange(0, min(results.num_display, graph.get_num_users())):
most_linked = graph.get_user_linked(i)
print str(i +
1) + "th most linked " + most_linked.name + " with " + str(
len(most_linked.followers))
graph.calculate_friends()
for i in xrange(0, min(results.num_display, graph.get_num_users())):
most_friends = graph.get_user_friends(i)
print str(i +
1) + "th most friends " + most_friends.name + " with " + str(
len(most_friends.friends))
print "Average Friends " + str(graph.avg_friends)
def init_kmeans(k, data):
print(data)
path = "data/" + data
S, S_ori, A, true_clus, flag, A_ori = build_graph.build_graph(path)
clus_list = list(set(true_clus))
print(clus_list)
clus_dic = {}
for i in range(len(clus_list)):
clus_dic[clus_list[i]] = i
for i in range(len(true_clus)):
true_clus[i] = clus_dic[true_clus[i]]
pred_l = []
cent_l = []
km_l = []
mod = []
ent = []
nmi = []
ari = []
for j in range(5):
pred, centroids, kmeans_clus = clustering(A_ori, k)
pred_l.append(pred)
cent_l.append(centroids)
km_l.append(kmeans_clus)
ari.append(evaluate.ARI(true_clus, pred))
ind = ari.index(sorted(ari)[2])
pred = pred_l[ind]
centroids = cent_l[ind]
kmeans_clus = km_l[ind]
U = initialize_U(A_ori, centroids)
f = open('initialize/' + data + '_U_' + str(k) + '.csv', 'w')
writer = csv.writer(f, lineterminator='\n')
writer.writerows(U)
f.close()
f = open('initialize/' + data + '_V_' + str(k) + '.csv', 'w')
writer = csv.writer(f, lineterminator='\n')
writer.writerows(centroids)
f.close()
flag=1
break
parent_id = child_ids(nn1_ind)
depth = depth+1
if flag == 1:
nn_id = parent_id
nn_dist = parent_dist
else:
nn_id = - 1
nn_dist = -1
return nn_id, nn_dist, visited
if __name__ == '__main__':
dt = numpy.dtype('f8')
X = [range(10)]
print X
for i in range(10):
Vect = [random.randint(0,100) for r in range(10)]
X = numpy.vstack([X,Vect])
X = numpy.delete(X,(0),axis = 0)
X = MA(X,dt)
K = 2
print X
nodes = build_graph(X,3)
Q = [random.randint(0,100) for r in range(10)]
print Q
Q = MA(Q,dt)
print (search_graph(Q,nodes,X,1))
def create_gaph(args):
ls_adj, feature_list, word_freq_list, y, y_hot, train_size, word_vectors, positions_list = build_graph.build_graph(
config=args.configfile)
vocab_size = len(word_vectors)
word_vectors = torch.from_numpy(word_vectors).float()
g_list = []
max_words = 0
for i, adj in enumerate(ls_adj):
adj = normalize_adj(adj)
g = nx.from_scipy_sparse_matrix(adj)
lb = y[i]
feat = feature_list[i]
recon = [0] * vocab_size
for j, f in enumerate(feat):
recon[f] = word_freq_list[i][0][j] * word_freq_list[i][1][j]
m_recon = max(recon)
for f in range(len(recon)):
recon[f] /= m_recon
# if frequency_as_feature:
# feat = np.concatenate((feat, word_freq_list[i].toarray()), axis=1)
# # feat = feat * word_freq_list[i].toarray()
if i == 10:
print(word_freq_list[i])
# s = sum(word_freq_list[i])
# # s = 1
# wf = [el / s for el in word_freq_list[i]]
s = sum(word_freq_list[i][0])
# s = 1
wf1 = [el / s for el in word_freq_list[i][0]]
s = sum(word_freq_list[i][1])
# s = 1
wf2 = [el / s for el in word_freq_list[i][1]]
wf = (wf1, wf2)
g_list.append(
S2VGraph(g,
lb,
node_features=feat,
word_freq=wf,
positions=positions_list[i],
adj=adj.todense(),
recon=recon))
for ar in positions_list[i]:
max_words = max(max_words, max(ar))
max_words += 1
zero_edges = 0
for g in g_list:
# edges = [list(pair) for pair in g.g.edges()]
# edges_w = [w['weight'] for i, j, w in g.g.edges(data=True)]
# edges.extend([[i, j] for j, i in edges])
# edges_w.extend([w for w in edges_w])
edges = []
edges_w = []
for i, j, wt in g.g.edges(data=True):
w = wt['weight']
edges.append([i, j])
edges_w.append(w)
if i != j:
edges.append([j, i])
edges_w.append(w)
if len(edges) == 0:
print('zero edge : ', len(g.g))
zero_edges += 1
edges = [[0, 0]]
edges_w = [0]
g.edge_mat = torch.tensor(edges).long().transpose(0, 1)
g.edges_weights = torch.tensor(edges_w).float()
print('total zero edge graphs : ', zero_edges)
return g_list, len(set(y)), train_size, word_vectors, max_words
FLAGS = flags.FLAGS
#args=getopt()
cv = 3
k = 6
fasta_name = 'first.txt'
data_name = fasta_name.split('.')[0]
split2cv(cv, fasta_name, data_name)
test_acc = []
test_pred = []
test_labels = []
for i in range(cv):
temp_data_name = data_name + '_cv' + str(i + 1)
print(temp_data_name)
prepare_data(temp_data_name, k)
build_graph(temp_data_name)
acc, pred, labels = train(temp_data_name)
test_acc.extend([acc])
test_labels.extend(labels)
test_pred.extend(pred)
print('cv_acc:', np.mean(np.array(test_acc)))
np.savetxt(data_name + '_cv_acc_result.csv',
np.array(test_acc),
delimiter=',',
fmt='%5f')
np.savetxt(data_name + 'cv_pred.csv',
np.array([test_labels, test_pred]).T,
delimiter=',',
fmt='%d')
test_data = 'PDB_test.txt'
test_label = 'test_label.txt'
# train_data = 'trainnew.txt'
# train_label = 'train_labelnew.txt'
# test_data = 'testnew.txt'
# test_label = 'test_labelnew.txt'
test_acc=[]
test_pred=[]
test_labels=[]
data_name=train_data.split('.')[0]
prepare_data_trian_test(train_data,train_label,test_data,test_label,data_name,k)
build_graph(data_name,word_embeddings_dim,slide_size)
acc,pred,labels=train(data_name)
test_acc.extend([acc])
test_labels.extend(labels)
test_pred.extend(pred)
print('cv_acc:',np.mean(np.array(test_acc)))
np.savetxt(data_name+'_cv_acc_result_test_'+str(k)+'.csv',np.array(test_acc),delimiter=',',fmt='%5f')
np.savetxt(data_name+'cv_pred_test_'+str(k)+'.csv',np.array([test_labels,test_pred]).T,delimiter=',',fmt='%d')
def build(self, reuse=False):
self.reuse = reuse
with tf.variable_scope(self.name, reuse=self.reuse):
# train set
self.mix = self.loader.mix_queue # mixture batch
self.source = self.loader.source_queue # source batch
self.other = self.mix - self.source
# test samples (placeholder)
self.mix_test = tf.placeholder(tf.float32, shape=(1,16384), name="mixture_sample")
self.source_test = tf.placeholder(tf.float32, shape=(1,16384), name="source_sample")
# define network
self.FE = FrontEnd()
if config.network == "unet":
self.SEP = Unet(model=config.model)
elif config.network == "lstm":
self.SEP = LSTM(model=config.model)
elif config.network == "bilstm":
self.SEP = BiLSTM(model=config.model)
elif config.network == "dilated":
self.SEP = Dilated(model=config.model)
else:
print("No model chosen")
raise ValueError
self.BE = BackEnd()
functions = [self.FE, self.SEP, self.BE]
if config.network2 == "unet":
self.SEP2 = Unet(model=config.model2)
self.CON = Confidence(model=config.confidence)
functions = functions + [self.SEP2, self.CON]
elif config.network2 == "lstm":
self.SEP2 = LSTM(model=config.model2)
self.CON = Confidence(model=config.confidence)
functions = functions + [self.SEP2, self.CON]
elif config.network2 == "bilstm":
self.SEP2 = BiLSTM(model=config.model2)
self.CON = Confidence(model=config.confidence)
functions = functions + [self.SEP2, self.CON]
elif config.network2 == "dilated":
self.SEP2 = Dilated(model=config.model2)
self.CON = Confidence(model=config.confidence)
functions = functions + [self.SEP2, self.CON]
else:
print("No model chosen")
input = {"mix": self.mix, "source": self.source}
input_test = {"mix": self.mix_test, "source": self.source_test}
# draw graph
self.graph = build_graph()
self.graph_test = build_graph(train=False)
self.graph(input, functions)
self.update_ops = tf.get_collection(tf.GraphKeys.UPDATE_OPS)
self.graph_test(input_test, functions)
# variable count
self.variable_size = np.sum(
np.array([np.prod(np.array(v.get_shape().as_list())) for v in tf.trainable_variables()]))
print("\n total varialbes : {}".format(self.variable_size), "\n")
for v in tf.trainable_variables():
print(v, "{}".format(np.prod(np.array(v.get_shape().as_list()))))
# Define loss and summarize
# self.loss_pre = (loss_fn(self.graph.masked_spec1, self.graph.estimated1, self.graph.source_spec,
# self.source, self.other, loss_type=config.loss) + \
# loss_fn(self.graph.masked_spec2, self.graph.estimated2, self.graph.source_spec,
# self.source, self.other, loss_type=config.loss))/2
if config.hybrid:
if config.loss_seq == 'mid':
self.loss = loss_fn(self.graph.masked_spec1_mid, self.graph.estimated1_mid, self.graph.source_spec,
self.source, self.other, loss_type=config.loss) + \
loss_fn(self.graph.masked_spec2_mid, self.graph.estimated2_mid, self.graph.source_spec,
self.source, self.other, loss_type=config.loss)
elif config.loss_seq == 'end':
self.loss = loss_fn(self.graph.masked_spec1, self.graph.estimated1, self.graph.source_spec,
self.source, self.other, loss_type=config.loss) + \
loss_fn(self.graph.masked_spec2, self.graph.estimated2, self.graph.source_spec,
self.source, self.other, loss_type=config.loss)
elif config.loss_seq == 'both':
self.loss = (loss_fn(self.graph.masked_spec1, self.graph.estimated1, self.graph.source_spec,
self.source, self.other, loss_type=config.loss) +
loss_fn(self.graph.masked_spec2, self.graph.estimated2, self.graph.source_spec,
self.source, self.other, loss_type=config.loss) +
loss_fn(self.graph.masked_spec1_mid, self.graph.estimated1_mid, self.graph.source_spec,
self.source, self.other, loss_type=config.loss) +
loss_fn(self.graph.masked_spec2_mid, self.graph.estimated2_mid, self.graph.source_spec,
self.source, self.other, loss_type=config.loss))/2
elif config.loss_seq == 'two':
self.loss_pre = loss_fn(self.graph.masked_spec1_mid, self.graph.estimated1_mid, self.graph.source_spec,
self.source, self.other, loss_type=config.loss) + \
loss_fn(self.graph.masked_spec2_mid, self.graph.estimated2_mid, self.graph.source_spec,
self.source, self.other, loss_type=config.loss)
self.loss = (loss_fn(self.graph.masked_spec1, self.graph.estimated1, self.graph.source_spec,
self.source, self.other, loss_type=config.loss) +
loss_fn(self.graph.masked_spec2, self.graph.estimated2, self.graph.source_spec,
self.source, self.other, loss_type=config.loss) +
loss_fn(self.graph.masked_spec1_mid, self.graph.estimated1_mid, self.graph.source_spec,
self.source, self.other, loss_type=config.loss) +
loss_fn(self.graph.masked_spec2_mid, self.graph.estimated2_mid, self.graph.source_spec,
self.source, self.other, loss_type=config.loss)) / 2
elif config.loss_seq == 'first':
self.loss_pre = loss_fn(self.graph.masked_spec1_mid, self.graph.estimated1_mid, self.graph.source_spec,
self.source, self.other, loss_type=config.loss)
self.loss = loss_fn(self.graph.masked_spec1, self.graph.estimated1, self.graph.source_spec,
self.source, self.other, loss_type=config.loss) + \
loss_fn(self.graph.masked_spec1_mid, self.graph.estimated1_mid, self.graph.source_spec,
self.source, self.other, loss_type=config.loss)
self.graph.estimated = self.graph.estimated1
self.graph_test.estimated = self.graph_test.estimated1
elif config.loss_seq == 'second':
self.loss_pre = loss_fn(self.graph.masked_spec2_mid, self.graph.estimated2_mid, self.graph.source_spec,
self.source, self.other, loss_type=config.loss)
self.loss = loss_fn(self.graph.masked_spec2, self.graph.estimated2, self.graph.source_spec,
self.source, self.other, loss_type=config.loss) + \
loss_fn(self.graph.masked_spec2_mid, self.graph.estimated2_mid, self.graph.source_spec,
self.source, self.other, loss_type=config.loss)
self.graph.estimated = self.graph.estimated2
self.graph_test.estimated = self.graph_test.estimated2
else:
raise AssertionError("wrong config.loss_seq !!")
else:
self.loss = loss_fn(self.graph.masked_spec, self.graph.estimated, self.graph.source_spec,
self.source, self.other, loss_type=config.loss)
self.loss_summary = tf.summary.scalar("loss", self.loss)
self.alpha_summary = tf.summary.scalar("alpha", tf.reduce_mean(self.graph.alpha))
self.summary = tf.summary.merge_all()
print("\n loss type : %s \n" % config.loss)
assert "visited_authors.pkl" in progress_dir, "visited_authors.pkl not found!" + prompt
assert args.max_book > 0, "max_book must be a positive integer."
assert args.max_book <= 2000, "max_book should be less than 2000."
assert args.max_author > 0, "max_author must be a positive integer."
assert args.max_author <= 2000, "max_author should be less than 2000."
if __name__ == "__main__":
parser = construct_parser()
validate_scrape_args(parser)
args = parser.parse_args()
if args.which == "scrape": # run command "scrape"
start_url = args.start_url
max_book = args.max_book
max_author = args.max_author
new_scrape = args.new
scrape_start(new_scrape, start_url, max_book, max_author, PROGRESS_DIR)
elif args.which == "update": # run command "update"
type_json = args.type
src_json = args.srcJSON
insert_into_db(src_json, type_json)
elif args.which == "export": # run command "export"
db_choice = args.db
dump_db(db_choice)
elif args.which == "draw": # run command "draw"
build_graph()
else: # invalid input
print("error : invalid input.")
parser.print_help()
import sys
from pprint import pprint
from build_graph import build_graph
from map_coloring_utils import mrv, degree_heuristic, lcv, get_allowed_colors, coloring
if __name__ == "__main__":
map = 'map.txt'
colors = ['RED', 'GREEN']
# Build graph from file
graph = build_graph(map)
for _ in range(len(graph)):
cities_with_max_degree = degree_heuristic(graph)
cities_with_minimum_remaining_colors = mrv(graph, colors)
much_used_colors = lcv(graph, colors)
selected_city = set(cities_with_max_degree).intersection(
set(cities_with_minimum_remaining_colors)).pop()
# Get allowed color for selected city
colors_of_selected_city = get_allowed_colors(graph, selected_city,
colors)
# Final chosen color
common_color = set(much_used_colors).intersection(
set(colors_of_selected_city))
try:
fp = 0
tn = 0
fn = 0
(graphs, test_edge_list) = self.graph.k_folds(k)
for graph, test_edges in zip(graphs, test_edge_list):
tc = TriadClassifier(graph)
tc.train()
for test in test_edges:
node1 = test[0][0]
node2 = test[0][1]
weight = test[1]
pred = tc.classify_pair(node1, node2)
if 1 * weight >= 0:
if pred[0] > pred[1]:
tp += 1
else:
fn += 1
else:
if pred[0] < pred[1]:
tn += 1
else:
fp += 1
return (tp, tn, fp, fn)
if __name__ == '__main__':
tc = TriadClassifier(build_graph.build_graph('data/mlb/2015'))
(tp, tn, fp, fn) = tc.k_folds()
print(tp, tn, fp, fn)
print "Accuracy: ", float(tp + tn) / (tp + fp + tn + fn)
def construct_spring_approximation(fname, verbose, run_tests, prune,load_dist, iterations_to_save = -1, solver = "BFGS", k = 1000, q = 500):
#Create the Graph
graph = build_graph.build_graph(fname,prune,verbose)
i = 0
j = 0
#unit distance
e = 1
#replusion distance scaler
scale = 1
position_matrix = e*graph.get_num_users()*np.random.random((graph.get_num_users(), 2))
length_matrix = np.zeros((graph.get_num_users(),graph.get_num_users()))
un_length_matrix = np.zeros((graph.get_num_users(),graph.get_num_users()))
if load_dist:
length_matrix = np.load("directed_length_matrix")
un_length_matrix = np.load("undirected_length_matrix")
else:
if verbose:
print "About to construct all pair distance matrices. This could take a while."
length_matrix = graph.compute_all_path_matrix(True)
un_length_matrix = graph.compute_all_path_matrix(False)
if verbose:
print "Done computing distance matrices"
print "Saving all pair distance matrices"
np.save("directed_length_matrix",length_matrix)
np.save("undirected_length_matrix",un_length_matrix)
connectivity_matrix = np.min(length_matrix,1)
connectivity_matrix[connectivity_matrix == 0] = 1
connectivity_matrix[connectivity_matrix == -1] = 0
repulsion_matrix = (1-connectivity_matrix)
print connectivity_matrix
print repulsion_matrix
print length_matrix
print un_length_matrix
#a quick vectorized distance function
dist = lambda P: np.sqrt((P[:,0]-P[:,0][:, np.newaxis])**2 +
(P[:,1]-P[:,1][:, np.newaxis])**2)
#general broadcasting function
def build_matrix(func, args):
return func(*args)
def diff(A,B):
return A[:,np.newaxis] - B
def energy(P):
epsilon = 0.000001
P = P.reshape((-1, 2))
D = dist(P)
# The potential energy is the sum of the elastic potential energies plus the repel energies.
return (k * connectivity_matrix * (D - length_matrix)**2).sum() + (q*(un_length_matrix)*(repulsion_matrix)*(1/(D/scale+epsilon))).sum()
def grad(P):
epsilon = 0.000001
P = P.reshape((-1, 2))
# We compute the distance matrix.
D = dist(P)
x_diff = build_matrix(diff,(P[:,0],P[:,0]))
y_diff = build_matrix(diff,(P[:,1],P[:,1]))
num_users = P.shape[0]
grad = np.zeros((num_users,2))
#Sorry there has to be a little math involved
grad[:,0]=((2*k*connectivity_matrix*(D-length_matrix)*(1/(D+epsilon)) - q*un_length_matrix*repulsion_matrix*1/((D/scale)**3+epsilon))*x_diff).sum(axis = 1)
grad[:,0]+=((2*k*np.transpose(connectivity_matrix)*(D-np.transpose(length_matrix))*(1/(D+epsilon)) - q*np.transpose(un_length_matrix)*np.transpose(repulsion_matrix)*1/((D/scale)**3+epsilon))*x_diff).sum(axis = 1)
grad[:,1]=((2*k*connectivity_matrix*(D-length_matrix)*(1/(D+epsilon)) - q*un_length_matrix*repulsion_matrix*1/((D/scale)**3+epsilon))*y_diff).sum(axis = 1)
grad[:,1]+=((2*k*np.transpose(connectivity_matrix)*(D-np.transpose(length_matrix))*(1/(D+epsilon)) - q*np.transpose(un_length_matrix)*np.transpose(repulsion_matrix)*1/((D/scale)**3+epsilon))*y_diff).sum(axis = 1)
return grad.ravel()
#Q:Why are there arrays here?
#A:Python's closures do not let you modify the variables outside of local scope but they do let you modify objects like arrays
num_iterations = [0]
best_soln = [None]
last_time = datetime.now()
def solver_callback(xk):
num_iterations.append(num_iterations[0] + 1)
num_iterations.pop(0)
best_soln.append(xk)
best_soln.pop(0)
print str(num_iterations[0]) + " iterations with energy " + str(energy(xk))
if num_iterations[0] % iterations_to_save == 0 and iterations_to_save > 0:
print str(iterations_to_save) + " iterations in:" + str((datetime.now()-last_time).seconds) + "seconds"
last_time = datetime.now()
print "Saving Current Solution..."
save_solution(graph,xk,"iteration_"+str(num_iterations[0]))
print "Solution Saved as " + "iteration_"+str(num_iterations[0])+".js"
def leave_handler(signal, frame):
if best_soln[0] != None:
print "Saving best solution on exit as exit_solution.js"
save_solution(graph, "exit_solution", "exit_solution")
return (solver_callback, leave_handler)
signal.signal(signal.SIGINT, leave_handler)
if run_tests:
from scipy.optimize import check_grad
from scipy.optimize import approx_fprime
print "Checking Gradient:"
print check_grad(energy, grad, position_matrix.ravel())
print "Speed Tests:"
print "Energy Function Speed:"
print datetime.now()
print energy(position_matrix.ravel())
print datetime.now()
print "Gradient Speed:"
print datetime.now()
print grad(position_matrix.ravel())
print datetime.now()
print "Minimizing with " + str(solver)
solution = opt.minimize(energy, position_matrix.ravel(),
method=solver,jac = grad,callback = solver_callback)
print "Solved to optimiality"
if verbose:
print energy(position_matrix.ravel())
print energy(solution.x)
save_solution(graph,solution.x,"results")
flag = 1
break
parent_id = child_ids(nn1_ind)
depth = depth + 1
if flag == 1:
nn_id = parent_id
nn_dist = parent_dist
else:
nn_id = -1
nn_dist = -1
return nn_id, nn_dist, visited
if __name__ == '__main__':
dt = numpy.dtype('f8')
X = [range(10)]
print X
for i in range(10):
Vect = [random.randint(0, 100) for r in range(10)]
X = numpy.vstack([X, Vect])
X = numpy.delete(X, (0), axis=0)
X = MA(X, dt)
K = 2
print X
nodes = build_graph(X, 3)
Q = [random.randint(0, 100) for r in range(10)]
print Q
Q = MA(Q, dt)
print(search_graph(Q, nodes, X, 1))
# https://www.reddit.com/r/dailyprogrammer/comments/4j65ls/20160513_challenge_266_hard_finding_friends_in/
import build_graph as bg
def maximum_clique(graph, vertices, clique=set()):
if not vertices:
return clique
return max((maximum_clique(graph, vertices & graph[v], clique | {v}) for v in vertices), key=len)
file_location = 'maximum_clique_input.txt'
print(maximum_clique(*bg.build_graph(file_location)))
def construct_spring_approximation(fname,
verbose,
run_tests,
prune,
load_dist,
iterations_to_save=-1,
solver="BFGS",
k=1000,
q=500):
#Create the Graph
graph = build_graph.build_graph(fname, prune, verbose)
i = 0
j = 0
#unit distance
e = 1
#replusion distance scaler
scale = 1
position_matrix = e * graph.get_num_users() * np.random.random(
(graph.get_num_users(), 2))
length_matrix = np.zeros((graph.get_num_users(), graph.get_num_users()))
un_length_matrix = np.zeros((graph.get_num_users(), graph.get_num_users()))
if load_dist:
length_matrix = np.load("directed_length_matrix")
un_length_matrix = np.load("undirected_length_matrix")
else:
if verbose:
print "About to construct all pair distance matrices. This could take a while."
length_matrix = graph.compute_all_path_matrix(True)
un_length_matrix = graph.compute_all_path_matrix(False)
if verbose:
print "Done computing distance matrices"
print "Saving all pair distance matrices"
np.save("directed_length_matrix", length_matrix)
np.save("undirected_length_matrix", un_length_matrix)
connectivity_matrix = np.min(length_matrix, 1)
connectivity_matrix[connectivity_matrix == 0] = 1
connectivity_matrix[connectivity_matrix == -1] = 0
repulsion_matrix = (1 - connectivity_matrix)
print connectivity_matrix
print repulsion_matrix
print length_matrix
print un_length_matrix
#a quick vectorized distance function
dist = lambda P: np.sqrt((P[:, 0] - P[:, 0][:, np.newaxis])**2 +
(P[:, 1] - P[:, 1][:, np.newaxis])**2)
#general broadcasting function
def build_matrix(func, args):
return func(*args)
def diff(A, B):
return A[:, np.newaxis] - B
def energy(P):
epsilon = 0.000001
P = P.reshape((-1, 2))
D = dist(P)
# The potential energy is the sum of the elastic potential energies plus the repel energies.
return (k * connectivity_matrix * (D - length_matrix)**2).sum() + (
q * (un_length_matrix) * (repulsion_matrix) *
(1 / (D / scale + epsilon))).sum()
def grad(P):
epsilon = 0.000001
P = P.reshape((-1, 2))
# We compute the distance matrix.
D = dist(P)
x_diff = build_matrix(diff, (P[:, 0], P[:, 0]))
y_diff = build_matrix(diff, (P[:, 1], P[:, 1]))
num_users = P.shape[0]
grad = np.zeros((num_users, 2))
#Sorry there has to be a little math involved
grad[:, 0] = (
(2 * k * connectivity_matrix * (D - length_matrix) *
(1 /
(D + epsilon)) - q * un_length_matrix * repulsion_matrix * 1 /
((D / scale)**3 + epsilon)) * x_diff).sum(axis=1)
grad[:,
0] += ((2 * k * np.transpose(connectivity_matrix) *
(D - np.transpose(length_matrix)) *
(1 / (D + epsilon)) - q * np.transpose(un_length_matrix) *
np.transpose(repulsion_matrix) * 1 /
((D / scale)**3 + epsilon)) * x_diff).sum(axis=1)
grad[:, 1] = (
(2 * k * connectivity_matrix * (D - length_matrix) *
(1 /
(D + epsilon)) - q * un_length_matrix * repulsion_matrix * 1 /
((D / scale)**3 + epsilon)) * y_diff).sum(axis=1)
grad[:,
1] += ((2 * k * np.transpose(connectivity_matrix) *
(D - np.transpose(length_matrix)) *
(1 / (D + epsilon)) - q * np.transpose(un_length_matrix) *
np.transpose(repulsion_matrix) * 1 /
((D / scale)**3 + epsilon)) * y_diff).sum(axis=1)
return grad.ravel()
#Q:Why are there arrays here?
#A:Python's closures do not let you modify the variables outside of local scope but they do let you modify objects like arrays
num_iterations = [0]
best_soln = [None]
last_time = datetime.now()
def solver_callback(xk):
num_iterations.append(num_iterations[0] + 1)
num_iterations.pop(0)
best_soln.append(xk)
best_soln.pop(0)
print str(num_iterations[0]) + " iterations with energy " + str(
energy(xk))
if num_iterations[
0] % iterations_to_save == 0 and iterations_to_save > 0:
print str(iterations_to_save) + " iterations in:" + str(
(datetime.now() - last_time).seconds) + "seconds"
last_time = datetime.now()
print "Saving Current Solution..."
save_solution(graph, xk, "iteration_" + str(num_iterations[0]))
print "Solution Saved as " + "iteration_" + str(
num_iterations[0]) + ".js"
def leave_handler(signal, frame):
if best_soln[0] != None:
print "Saving best solution on exit as exit_solution.js"
save_solution(graph, "exit_solution", "exit_solution")
return (solver_callback, leave_handler)
signal.signal(signal.SIGINT, leave_handler)
if run_tests:
from scipy.optimize import check_grad
from scipy.optimize import approx_fprime
print "Checking Gradient:"
print check_grad(energy, grad, position_matrix.ravel())
print "Speed Tests:"
print "Energy Function Speed:"
print datetime.now()
print energy(position_matrix.ravel())
print datetime.now()
print "Gradient Speed:"
print datetime.now()
print grad(position_matrix.ravel())
print datetime.now()
print "Minimizing with " + str(solver)
solution = opt.minimize(energy,
position_matrix.ravel(),
method=solver,
jac=grad,
callback=solver_callback)
print "Solved to optimiality"
if verbose:
print energy(position_matrix.ravel())
print energy(solution.x)
save_solution(graph, solution.x, "results")
def main(start = None, end = None):
#list of strings containing every street with an
#intersection in the grid
exp_node_num = 31
exp_edge_num = 47
streets_in_grid = ["SACRAMENTO ST", "BAKER ST", "SUTTER ST", "STEINER ST", "PERINE PL", "WILMOT ST", "BRODERICK ST", "DIVISADERO ST", "SCOTT ST", "PIERCE ST", "CALIFORNIA ST", "PINE ST", "BUSH ST"]
int_dic = loadIntersections(toPrint=False)
ints_in_grid = {}
#if both streets for intersection are in grid, add the intersection to ints_in_grid
for key in int_dic.keys():
if int_dic[key].get_streets()[0] in streets_in_grid and int_dic[key].get_streets()[1] in streets_in_grid:
ints_in_grid[key] = int_dic[key]
# for key in ints_in_grid.keys():
# print(ints_in_grid[key].get_streets())
# print(len(ints_in_grid))
#get every street where start int and end int are in test Grid
all_streets = load_streets(toPrint=False)
streets_in_grid = {}
for key in all_streets.keys():
if all_streets[key].get_start() in ints_in_grid.keys() and all_streets[key].get_end() in ints_in_grid.keys():
streets_in_grid[key] = all_streets[key]
# for key in streets_in_grid.keys():
# sne = streets_in_grid[key].get_start_name_end()
# print("{}->{}->{}".format(ints_in_grid[sne[0]].get_streets(), sne[1], ints_in_grid[sne[2]].get_streets()))
# print(len(streets_in_grid))
digraph = build_graph(ints_in_grid, streets_in_grid)
# print("Args: {}".format(sys.argv))
# print("Arg length: {}".format(len(sys.argv)))
if len(sys.argv) == 3:
if sys.argv[1] in ints_in_grid.keys() and sys.argv[2] in ints_in_grid.keys():
start = sys.argv[1]
end = sys.argv[2]
else:
print("invalid start or end node")
assert False
else:
start = random.choice(list(ints_in_grid.keys()))
end = random.choice(list(ints_in_grid.keys()))
while end == start:
end = random.choice(list(ints_in_grid.keys()))
start_int = ints_in_grid[start]
end_int = ints_in_grid[end]
start_node= SF_intersection(start_int.get_cnn(), start_int.get_streets(), start_int.get_loc(), start_int.get_elev())
end_node= SF_intersection(end_int.get_cnn(), end_int.get_streets(), end_int.get_loc(), end_int.get_elev())
bf_path = search(digraph, start_node, end_node)
print("start:{} - {}".format(start_int.get_streets(), start_int.get_cnn()))
print("end:{} - {}".format(end_int.get_streets(), end_int.get_cnn()))
print("brute force: " + str(bf_path))
print(bf_path.get_weight())
print()
dp_path = dsearch(digraph, start_node, end_node)
print()
print("dynamic p: " + str(dp_path))
print(dp_path.get_node_list())
print(dp_path.get_weight())
print("\nDystra:")
dpath = dijkstra_search(digraph, start_node, end_node, toPrint=False)
dyspath = makePathFromDijk(digraph, dpath)
print(dyspath)
print(dpath)
print(dyspath.get_weight())
dist = bfs(graph, vertices, u)
dist = [dist[v] for v in dist.keys() if v != u and dist[v] != INFINITY]
if dist: return max(dist)
else: return None
def bfs(graph, vertices, s):
dist = {}
for u in vertices:
dist[u] = INFINITY
dist[s] = 0
q = [s]
while q:
u = q.pop(0)
if not u in graph:
continue
for v in graph[u]:
if dist[v] > dist[u] + 1:
dist[v] = dist[u] + 1
q.append(v)
return dist
file_location = 'graph_size_input.txt'
ecc = eccentricities(*bg.build_graph(file_location, True))
radius = min(ecc)
diameter = max(ecc)
print('radius: %i\ndiameter: %i' % (radius, diameter))
def __init__(self,
hparams,
mode,
iterator,
handle,
vocab_table,
reverse_vocab_table=None,
scope=None,
extra_args=None):
assert isinstance(iterator, iterator_utils.BatchedInput)
self.iterator = iterator
self.handle = handle
self.mode = mode
self.vocab_table = vocab_table
self.vocab_size = hparams.vocab_size
self.num_layers = hparams.num_layers
self.num_gpus = hparams.num_gpus
self.hparams = hparams
self.single_cell_fn = None
self.global_gpu_num = 0
if extra_args:
self.single_cell_fn = extra_args.single_cell_fn
# Initializer
initializer = model_helper.get_initializer(hparams.init_op,
hparams.random_seed,
hparams.init_weight)
tf.get_variable_scope().set_initializer(initializer)
# Embeddings
self.init_embeddings(hparams, scope)
self.batch_size = tf.shape(self.iterator.source)[0]
# Projection
with tf.variable_scope(scope or "build_network"):
with tf.variable_scope("decoder/output_projection"):
self.output_layer1 = layers_core.Dense(
hparams.vocab_size,
use_bias=False,
name="output_projection_1")
self.output_layer2 = layers_core.Dense(
hparams.vocab_size,
use_bias=False,
name="output_projection_2")
self.output_layer_action = layers_core.Dense(
hparams.vocab_size,
use_bias=False,
name="output_projection_action")
self.vn_project11 = layers_core.Dense(
hparams.unit_value_network,
use_bias=False,
name="vn_project_11")
self.vn_project12 = layers_core.Dense(
hparams.unit_value_network,
use_bias=False,
name="vn_project_12")
self.vn_project21 = layers_core.Dense(
hparams.unit_value_network,
use_bias=False,
name="vn_project_21")
self.vn_project22 = layers_core.Dense(
hparams.unit_value_network,
use_bias=False,
name="vn_project_22")
## Train graph
sl_loss, sl_loss_arr, rl_loss_arr, sample_id_arr_train, sample_id_arr_infer = build_graph(
self, hparams, scope=scope)
if self.mode == tf.contrib.learn.ModeKeys.TRAIN:
self.train_loss = sl_loss
self.all_train_loss = sl_loss_arr
self.word_count = tf.reduce_sum(self.iterator.dialogue_len)
self.sample_ids_arr = sample_id_arr_train
self.sample_words_arr1 = []
self.sample_words_arr2 = []
source = self.iterator.source
for i in range(len(self.sample_ids_arr)):
element_infer = self.sample_ids_arr[i]
element_src = source[0]
# element_src=0
src = reverse_vocab_table.lookup(tf.to_int64(element_src))
infer = reverse_vocab_table.lookup(
tf.to_int64(element_infer)
)[0] # src can only get the first one so I only get the first inference
if i == 0:
self.sample_words_arr1.append((tf.constant(i), src, infer))
elif i == 1:
self.sample_words_arr2.append((tf.constant(i), src, infer))
self.vl1, self.vl2, self.pl1, self.pl2, self.eq11, self.eq12, self.eq2 = rl_loss_arr # reinforcement updates
elif self.mode == tf.contrib.learn.ModeKeys.EVAL:
self.eval_loss = sl_loss
self.all_eval_loss = sl_loss_arr
elif self.mode == tf.contrib.learn.ModeKeys.INFER:
self.sample_ids_arr = sample_id_arr_infer
self.sample_words_arr = []
self.source = reverse_vocab_table.lookup(
tf.to_int64(iterator.source))
for element in self.sample_ids_arr:
self.sample_words_arr.append(
reverse_vocab_table.lookup(tf.to_int64(element)))
elif self.mode in dialogue_utils.self_play_modes:
#### self play
self.train_loss = sl_loss
self.all_train_loss = sl_loss_arr
self.selfplay_agent_1_utt = reverse_vocab_table.lookup(
tf.to_int64(sample_id_arr_infer[0]))
self.selfplay_agent_2_utt = reverse_vocab_table.lookup(
tf.to_int64(sample_id_arr_infer[1]))
self.selfplay_action = reverse_vocab_table.lookup(
tf.to_int64(sample_id_arr_infer[2]))
if self.mode == dialogue_utils.mode_self_play_mutable:
self.vl1, self.vl2, self.pl1, self.pl2, self.eq11, self.eq12, self.eq2 = rl_loss_arr # reinforcement updates
if self.mode != tf.contrib.learn.ModeKeys.INFER:
## Count the number of predicted words for compute ppl.
self.predict_count = tf.reduce_sum(self.iterator.dialogue_len)
## Learning rate
warmup_steps = hparams.learning_rate_warmup_steps
warmup_factor = hparams.learning_rate_warmup_factor
print(" start_decay_step=%d, learning_rate=%g, decay_steps %d, "
"decay_factor %g, learning_rate_warmup_steps=%d, "
"learning_rate_warmup_factor=%g, starting_learning_rate=%g" %
(hparams.start_decay_step, hparams.learning_rate,
hparams.decay_steps, hparams.decay_factor, warmup_steps,
warmup_factor,
(hparams.learning_rate * warmup_factor**warmup_steps)))
self.global_step = tf.Variable(0, trainable=False)
params = tf.trainable_variables()
# Gradients and SGD update operation for training the model.
# Arrage for the embedding vars to appear at the beginning.
if self.mode == tf.contrib.learn.ModeKeys.TRAIN or self.mode == dialogue_utils.mode_self_play_mutable:
self.learning_rate = tf.constant(hparams.learning_rate)
inv_decay = warmup_factor**(tf.to_float(warmup_steps -
self.global_step))
self.learning_rate = tf.cond(
self.global_step < hparams.learning_rate_warmup_steps,
lambda: inv_decay * self.learning_rate,
lambda: self.learning_rate,
name="learning_rate_decay_warump_cond")
if hparams.optimizer == "sgd":
self.learning_rate = tf.cond(
self.global_step < hparams.start_decay_step,
lambda: self.learning_rate,
lambda: tf.train.exponential_decay(self.learning_rate, (
self.global_step - hparams.start_decay_step),
hparams.decay_steps,
hparams.decay_factor,
staircase=True),
name="sgd_learning_rate_supervised")
opt = tf.train.GradientDescentOptimizer(self.learning_rate,
name="SGD_supervised")
tf.summary.scalar("lr", self.learning_rate)
elif hparams.optimizer == "adam":
assert float(
hparams.learning_rate
) <= 0.001, "! High Adam learning rate %g" % hparams.learning_rate
opt = tf.train.AdamOptimizer(self.learning_rate,
name="Adam_supervised")
gradients = tf.gradients(self.train_loss,
params,
colocate_gradients_with_ops=hparams.
colocate_gradients_with_ops,
name="gradients_adam")
clipped_gradients, gradient_norm_summary = model_helper.gradient_clip(
gradients, max_gradient_norm=hparams.max_gradient_norm)
self.update = opt.apply_gradients(zip(clipped_gradients, params),
global_step=self.global_step,
name="adam_apply_gradients")
# Summary
self.train_summary = tf.summary.merge([
tf.summary.scalar("lr", self.learning_rate),
tf.summary.scalar("train_loss", self.train_loss),
] + gradient_norm_summary)
# second part of the learning rate
if self.mode == tf.contrib.learn.ModeKeys.TRAIN or self.mode == dialogue_utils.mode_self_play_mutable:
self.learning_rate2 = tf.constant(hparams.learning_rate2)
self.learning_rate3 = tf.constant(hparams.learning_rate3)
if hparams.optimizer == "sgd":
self.learning_rate2 = tf.cond(
self.global_step < hparams.start_decay_step,
lambda: self.learning_rate2,
lambda: tf.train.exponential_decay(self.learning_rate2, (
self.global_step - hparams.start_decay_step),
hparams.decay_steps,
hparams.decay_factor,
staircase=True),
name="sgd_learning_rate_supervised2")
self.learning_rate3 = tf.cond(
self.global_step < hparams.start_decay_step,
lambda: self.learning_rate3,
lambda: tf.train.exponential_decay(self.learning_rate3, (
self.global_step - hparams.start_decay_step),
hparams.decay_steps,
hparams.decay_factor,
staircase=True),
name="sgd_learning_rate_supervised3")
tf.summary.scalar("self_play_lr", self.learning_rate)
elif hparams.optimizer == "adam":
assert float(
hparams.learning_rate2
) <= 0.001, "! High Adam learning rate2 %g" % hparams.learning_rate2
assert float(
hparams.learning_rate3
) <= 0.001, "! High Adam learning rate3 %g" % hparams.learning_rate3
# params=[]
print("params=")
for element in params:
print(element.name)
val1_params = self.patial_params(
params, ["dynamic_seq2seq/value_network1"])
val2_params = self.patial_params(
params, ["dynamic_seq2seq/value_network2"])
embedding_params = self.patial_params(params, ["embeddings"])
main_dec_enc_params1 = self.patial_params(
params,
["dynamic_seq2seq/encoder1/", "dynamic_seq2seq/decoder1/"])
main_dec_enc_params2 = self.patial_params(
params,
["dynamic_seq2seq/encoder2/", "dynamic_seq2seq/decoder2/"])
action_params = self.patial_params(
params, ["dynamic_seq2seq/decoder_action"])
encoder_kb_params = self.patial_params(
params, ["dynamic_seq2seq/encoder2_kb"])
encoder_intent_params = self.patial_params(
params, ["dynamic_seq2seq/encoder1_intent"])
print("val1_params", "\n".join(map(lambda a: a.name, val1_params)))
print("val2_params", "\n".join(map(lambda a: a.name, val2_params)))
print("embedding_params",
"\n".join(map(lambda a: a.name, embedding_params)))
print("main_dec_enc_params1",
"\n".join(map(lambda a: a.name, main_dec_enc_params1)))
print("main_dec_enc_params2",
"\n".join(map(lambda a: a.name, main_dec_enc_params2)))
print("action_params",
"\n".join(map(lambda a: a.name, action_params)))
print("encoder_kb_params",
"\n".join(map(lambda a: a.name, encoder_kb_params)))
print("encoder_intent_params",
"\n".join(map(lambda a: a.name, encoder_intent_params)))
self.optimizer_vl1, self.v1_sum = self.generate_optimizer(
self.vl1, params, "vl1", self.learning_rate2,
self.hparams.max_gradient_norm2)
self.optimizer_vl2, self.v2_sum = self.generate_optimizer(
self.vl2, params, "vl2", self.learning_rate2,
self.hparams.max_gradient_norm2)
if hparams.self_play_variable_method == 0:
rl_param1, rl_param2 = encoder_intent_params, encoder_kb_params + action_params
elif hparams.self_play_variable_method == 1:
rl_param1, rl_param2 = main_dec_enc_params1, main_dec_enc_params2
elif hparams.self_play_variable_method == 2:
rl_param1, rl_param2 = main_dec_enc_params1 + encoder_intent_params, main_dec_enc_params2 + encoder_kb_params + action_params
elif hparams.self_play_variable_method == 3:
rl_param1, rl_param2 = [main_dec_enc_params1[0]
] + encoder_intent_params, [
main_dec_enc_params2[0]
] + encoder_kb_params
elif hparams.self_play_variable_method == 4:
rl_param1, rl_param2 = [main_dec_enc_params1[0]
], [main_dec_enc_params2[0]]
elif hparams.self_play_variable_method == 5:
rl_param1, rl_param2 = params, params
self.optimizer_pl1, self.p1_sum = self.generate_optimizer(
self.pl1, params, "pl1", self.learning_rate3,
self.hparams.max_gradient_norm3)
self.optimizer_pl2, self.p2_sum = self.generate_optimizer(
self.pl2, params, "pl2", self.learning_rate3,
self.hparams.max_gradient_norm3)
print("self.learning", self.learning_rate, self.learning_rate2,
self.learning_rate3)
################################
### supervised learning######'
###########################
# Saver
self.saver = tf.train.Saver(tf.global_variables())
# Print trainable variables
utils.print_out("# Trainable variables")
for param in params:
utils.print_out(
" %s, %s, %s" %
(param.name, str(param.get_shape()), param.op.device))
"""
from bigquery import user_input_query_helper
from build_graph import build_graph
from cycles import transaction_cycle
from high_balance import find_avg_balance, high_balance_transactions
from subnetworks import biggest_subnetwork, future_partners
from regression import balance_correlation_and_plot
from visualize_graph import plot_graph
# Prompt user for input, run a query on BigQuery, and write the results to csv files.
# If the query fails due to exceeding the quota, replace
# 'credentials.json' with 'backup-credentials.json'
# user_input_query_helper('credentials.json')
# Create a graph of the ethereum network using the csv files available.
ethereum_graph = build_graph('balances.csv', 'transactions.csv')
# Visualize the graph
plot_graph(ethereum_graph)
# Run the linear regression and output the result.
r2, rmse = balance_correlation_and_plot(ethereum_graph)
print("Coefficient of determination (r^2): " + str(r2) + "\nRMSE: " +
str(rmse) + "\n")
# Prompt the user if they are ready to run high_balance, run it if they are.
print("Enter 'y' when you wish to run the high_balance.py.")
user_input = input("Are you ready?: ")
if user_input.lower() == 'y':
avg = find_avg_balance(ethereum_graph)
args=parse.parse_args()
return args
'''
if __name__ == "__main__":
flags = tf.app.flags
FLAGS = flags.FLAGS
#args=getopt()
k=9
train_data='train.data'
train_label='train.label'
test_data='test.data'
test_label='test.label'
data_name=train_data.split('.')[0]
prepare_data_trian_test(train_data,train_label,test_data,test_label,data_name,k)
build_graph(data_name)
train(data_name)
return x['startIndex'] < y['startIndex']
if x['endIndex'] != y['endIndex']:
return x['endIndex'] < y['endIndex']
if not os.path.exists('results'):
os.mkdir('results')
# filenames = ['Anatole France.json']
# filenames = ['Barack Obama.json']
vocab = ['his', 'her']
filenames = os.listdir('cluster')
filenames.remove('Canada.json')
for filename in filenames:
print('File: {}'.format(filename))
graph = build_graph(filename)
with open(os.path.join('cluster', filename)) as f:
clusters = json.load(f)
clusters_not_assigned = []
for cluster in clusters:
cluster.sort(
key=lambda x: (x['sentNum'], x['startIndex'], x['endIndex']))
node = None
for mention_idx, mention in enumerate(cluster):
node_, max_len = search_node(graph, mention)
if node_ is not None:
if node is None:
node_.addMentions(cluster[:mention_idx])
node = node_
def train(num_epochs, beta, lr, evaluate_test_dset=False):
tf.logging.set_verbosity(tf.logging.INFO)
tf.reset_default_graph()
inputs, outputs, savers, tf_summaries = build_graph(beta=beta)
learning_rate, rgb_input, flow_input, is_training, y = inputs
scores, loss, loss_minimize = outputs
rgb_saver, flow_saver, training_saver = savers
# Load the training and test data
X_train_initial, X_test, y_train_initial, y_test = load_exercise_dataset(_VIDEO_DIR, _LABEL_MAP_PATH)
X_train_initial, y_train_initial = np.array(X_train_initial), np.array(y_train_initial)
dset_size = X_train_initial.shape[0]
validation_cutoff = int(dset_size * 0.2)
test_dset = list(zip(X_test, y_test))
print('\ntrain_dset len={}; val_dset len={}; test_dset len={}\n'.format(
dset_size - validation_cutoff, validation_cutoff, len(test_dset)
))
try:
train_accuracies = np.load(_STATS['train_acc']).tolist()
val_accuracies = np.load(_STATS['val_acc']).tolist()
losses = np.load(_STATS['loss']).tolist()
tf.logging.info('Statistics restored')
except:
train_accuracies, val_accuracies, losses = [], [], []
def _check_acc(msg, dset, sess):
num_correct, num_samples = 0, 0
for x_video, y_class in dset:
feed_dict = {
rgb_input: rgb_data(x_video, IMAGE_SIZE, nframes=NUM_FRAMES),
flow_input: flow_data(x_video, IMAGE_SIZE, nframes=NUM_FRAMES),
is_training: 0
}
scores_np = sess.run(scores, feed_dict=feed_dict)
y_pred = scores_np.argmax(axis=1)
num_samples += 1
num_correct += (y_pred == y_class).sum()
acc = float(num_correct) / num_samples
print('%s: %d / %d correct (%.2f%%)' % (msg, num_correct, num_samples, 100 * acc))
return acc
if not os.path.exists('summaries'):
os.mkdir('summaries')
path = os.path.join('summaries', 'first')
if not os.path.exists(path):
os.mkdir(path)
# Now we can run the computational graph many times to train the model.
# When we call sess.run we ask it to evaluate train_op, which causes the
# model to update.
with tf.Session() as sess:
writer = tf.summary.FileWriter(path, sess.graph)
sess.run(tf.global_variables_initializer())
rgb_saver.restore(sess, _CHECKPOINT_PATHS['rgb_imagenet'])
tf.logging.info('RGB checkpoint restored')
flow_saver.restore(sess, _CHECKPOINT_PATHS['flow_imagenet'])
tf.logging.info('Flow checkpoint restored')
try:
training_saver.restore(sess, _CHECKPOINT_PATHS['training'])
tf.logging.info('Training checkpoint restored')
except Exception as e:
pass
if evaluate_test_dset:
_ = _check_acc('Test', test_dset, sess)
exit()
t = 0
for epoch in range(num_epochs):
print('Starting epoch %d' % epoch)
# Re-sample train and validation datasets for each epoch
nums = list(range(dset_size))
indices = random.sample(nums, validation_cutoff)
mask = np.ones(dset_size, np.bool)
mask[indices] = 0
X_train, y_train = X_train_initial[mask], y_train_initial[mask]
train_dset = list(zip(X_train, y_train))
X_val, y_val = X_train_initial[indices], y_train_initial[indices]
val_dset = list(zip(X_val, y_val))
if epoch != 0:
# Check training and validation accuracies, and save the model
save_path = training_saver.save(sess, _CHECKPOINT_PATHS['training'])
print('\nTraining model saved in path: %s' % save_path)
train_acc = _check_acc('Train', train_dset, sess)
val_acc = _check_acc('Val', val_dset, sess)
train_accuracies.append(train_acc)
val_accuracies.append(val_acc)
np.save(_STATS['train_acc'], np.array(train_accuracies))
np.save(_STATS['val_acc'], np.array(val_accuracies))
np.save(_STATS['loss'], np.array(losses))
for x_video, y_class in train_dset:
feed_dict = {
learning_rate: lr,
rgb_input: rgb_data(x_video, IMAGE_SIZE, nframes=NUM_FRAMES),
flow_input: flow_data(x_video, IMAGE_SIZE, nframes=NUM_FRAMES),
y: np.array([y_class]),
is_training: 1
}
if t % _CHECK_EVERY == 0:
ops = [loss, loss_minimize, tf_summaries]
loss_np, _, summary = sess.run(ops, feed_dict=feed_dict)
writer.add_summary(summary, epoch)
else:
ops = [loss, loss_minimize]
loss_np, _ = sess.run(ops, feed_dict=feed_dict)
losses.append(loss_np)
print('Iteration %d, loss = %.4f' % (t, loss_np))
t += 1
border_cells = {
cell
for cell in adj_cells if cell not in live_cells
}
new_border_cells = new_border_cells.union(border_cells)
# set flag
for cell in new_border_cells:
cells[str(cell)]['border'] = True
# write annual data
with open('../resources/' + year + '_cell_data.json', 'w') as f:
f.write(
json.dumps(cells, indent=4, sort_keys=True, separators=(',', ':')))
# for all years add layers until the graph is connected
for y in range(1985, 2017):
year = str(y)
print('Connecting graph for %s...' % year)
G = build_graph(year)
layer = 1
while not networkx.is_connected(G):
print('\tAdding border layer #%d...' % layer)
add_layer(year)
G = build_graph(year)
layer += 1
# EOF
organic_search_volume,
seo_value,
]]
df = pd.read_csv('data.csv')
scaler = StandardScaler()
features = [
'Analyst value (0 - 5)', 'Partner value (0 - 5)',
'Persona value (0 - 5)', 'Growing market', 'Organic Search Volume',
'SEO Value (0 - 3)'
]
scaler.fit(df[features])
y_scaled = scaler.transform(y)
loaded_model = pickle.load(open('model/gm_for_predicton_6dims.sav', 'rb'))
result = loaded_model.predict(y)[0]
build_graph(y_scaled)
new_graph = True
responses = [
"0>>> Not so sure about this one 😐️",
"1>>> It’s a 🦄! This api has a good chance of increasing traffic.",
"2>>> This one is probably not going to do so well🥶",
]
if result in [0, 1, 2]:
st.markdown(f"<p style='text-align: center'>Result:<br>\
{responses[result]}</p>",
unsafe_allow_html=True)
if new_graph:
st.image('pic.png', use_column_width=True, output_format='PNG')
else:
#args=getopt()
cv = 5
k = 3
fasta_name = 'PDB14120.txt'
#fasta_name='train.txt'
data_name = fasta_name.split('.')[0]
split2cv(cv, fasta_name, data_name)
test_acc = []
test_pred = []
test_labels = []
for i in range(cv):
temp_data_name = data_name + '_cv' + str(i + 1)
print(temp_data_name)
prepare_data(temp_data_name, k)
build_graph(temp_data_name, 20, 20)
acc, pred, labels = train(temp_data_name)
test_acc.extend([acc])
test_labels.extend(labels)
test_pred.extend(pred)
print('cv_acc:', np.mean(np.array(test_acc)))
np.savetxt(data_name + '_cv_acc_result.csv',
np.array(test_acc),
delimiter=',',
fmt='%5f')
np.savetxt(data_name + 'cv_pred.csv',
np.array([test_labels, test_pred]).T,
delimiter=',',
fmt='%d')