def get_rbn_meta(i):
print "GOGO" + str(i)
rbn_reservoir = RBNNode(connectivity=conf.connectivity,
n_nodes=n_nodes,
input_connectivity=n_nodes / 2)
reservoir_system = ReservoirSystem(rbn_reservoir)
reservoir_system.train_on(*conf.training_data)
accuracy = reservoir_system.test_on(*conf.test_data)
bns_file = "{}/{}.bns".format(conf.output_dir, i)
cnet_file = "{}/{}.cnet".format(conf.output_dir, i)
acc_file = "{}/{}.acc".format(conf.output_dir, i)
with open(cnet_file, 'w') as f:
f.write(rbn_reservoir.into_cnet())
with open(bns_file, 'w') as f:
subprocess.call(['./analytics/SAT/bns', cnet_file], stdout=f)
with open(bns_file, 'r') as bns:
attractors = map(
int, re.findall("Attractor \d+ is of length (\d+)",
bns.read()))
with open(acc_file, 'w') as acc:
acc.write(
json.dumps({
"accuracy": accuracy,
"attractors": attractors,
"mean_attractor_length": np.mean(attractors),
"n_attractors": len(attractors)
}))
def get_rbn_meta(i):
print "GOGO" + str(i)
rbn_reservoir = RBNNode(connectivity=conf.connectivity,
n_nodes=n_nodes,
input_connectivity=n_nodes/2)
reservoir_system = ReservoirSystem(rbn_reservoir)
reservoir_system.train_on(*conf.training_data)
accuracy = reservoir_system.test_on(*conf.test_data)
bns_file = "{}/{}.bns".format(conf.output_dir, i)
cnet_file = "{}/{}.cnet".format(conf.output_dir, i)
acc_file = "{}/{}.acc".format(conf.output_dir, i)
with open(cnet_file, 'w') as f:
f.write(rbn_reservoir.into_cnet())
with open(bns_file, 'w') as f:
subprocess.call(['./analytics/SAT/bns', cnet_file], stdout=f)
with open(bns_file, 'r') as bns:
attractors = map(int, re.findall("Attractor \d+ is of length (\d+)", bns.read()))
with open(acc_file, 'w') as acc:
acc.write(json.dumps({
"accuracy": accuracy,
"attractors": attractors,
"mean_attractor_length": np.mean(attractors),
"n_attractors": len(attractors)
}))
def calculate_accuracy((n_nodes, input_connectivity, output_connectivity)):
accuracies = []
description = '-'.join(
map(str, [n_nodes, input_connectivity, output_connectivity]))
print "START-" + description
for sample in range(conf.n_samples):
rbn_reservoir = RBNNode(connectivity=conf.connectivity,
input_connectivity=input_connectivity,
n_nodes=n_nodes,
output_connectivity=output_connectivity)
reservoir_system = ReservoirSystem(rbn_reservoir)
reservoir_system.train_on(*conf.training_data)
accuracy = reservoir_system.test_on(*conf.test_data)
accuracies.append(accuracy)
print "END-" + description
return {
"accuracies": accuracies,
"n_nodes": n_nodes,
"n_samples": conf.n_samples,
"input_connectivity": input_connectivity,
"output_connectivity": output_connectivity
}
def get_rbn_meta(_):
print "!!!", _
rbn_reservoir = RBNNode(connectivity=conf.connectivity,
input_connectivity=reservoir_size / 2,
n_nodes=reservoir_size)
reservoir_system = ReservoirSystem(rbn_reservoir)
reservoir_system.train_on(*conf.training_data)
accuracy = reservoir_system.test_on(*conf.test_data)
attractors_meta = find_attractors(rbn_reservoir)
attractors_meta["accuracy"] = accuracy
return attractors_meta
from tasks.temporal import create_datasets
from rbn.sklearn_rbn_node import RBNNode
from sklearn.linear_model import Ridge
datasets = create_datasets(10,
task_size=200,
window_size=3,
dataset_type='temporal_parity')
training_datasets, test_dataset = datasets[:-1], datasets[-1]
rbn_reservoir = RBNNode(connectivity=3,
input_connectivity=50,
n_nodes=100,
output_connectivity=0)
readout_layer = Ridge()
for td in training_datasets[:1]:
#rbn_reservoir.reset_state()
reservoir_input, expected_output = td
states = rbn_reservoir.execute(reservoir_input)
readout_layer.fit(states, expected_output)
predictions = readout_layer.predict(states)
for prediction in predictions:
prediction[0] = 1 if prediction[0] > 0.5 else 0
errors = sum(predictions != expected_output)
accuracy = 1 - float(errors) / len(predictions)
from tasks.temporal import create_datasets
from rbn.sklearn_rbn_node import RBNNode
from sklearn.linear_model import Ridge
datasets = create_datasets(
10,
task_size=200,
window_size=3,
dataset_type='temporal_parity')
training_datasets, test_dataset = datasets[:-1], datasets[-1]
rbn_reservoir = RBNNode(connectivity=3,
input_connectivity=50,
n_nodes=100,
output_connectivity=0)
readout_layer = Ridge()
for td in training_datasets[:1]:
rbn_reservoir.reset_state()
reservoir_input, expected_output = td
states = rbn_reservoir.execute(reservoir_input)
readout_layer.fit(states, expected_output)
predictions = readout_layer.predict(states)
for prediction in predictions:
prediction[0] = 1 if prediction[0] > 0.5 else 0
errors = sum(predictions != expected_output)
accuracy = 1 - float(errors) / len(predictions)