def test_SimpleESN_transform():
esn = SimpleESN(n_readout=n_readout)
echoes = esn.transform(X)
assert_true(esn.weights_ is not None)
assert_true(esn.input_weights_ is not None)
assert_true(esn.readout_idx_ is not None)
repeated_echoes = esn.transform(X)
assert_array_equal(echoes, repeated_echoes)
def test_SimpleESN_initialization():
esn = SimpleESN(n_readout,
n_components=100,
damping=0.5,
weight_scaling=0.9,
discard_steps=0,
random_state=None)
echoes = esn.fit_transform(X)
assert_true(echoes.shape == (n_samples, n_readout))
import numpy as np
if __name__ == '__main__':
X = loadtxt('MackeyGlass_t17.txt')
X = atleast_2d(X).T
train_length = 2000
test_length = 2000
X_train = X[:train_length]
y_train = X[1:train_length + 1]
X_test = X[train_length:train_length + test_length]
y_test = X[train_length + 1:train_length + test_length + 1]
# Simple training
my_esn = SimpleESN(n_readout=1000,
n_components=1000,
damping=0.3,
weight_scaling=1.25)
echo_train = my_esn.fit_transform(X_train)
regr = Ridge(alpha=0.01)
regr.fit(echo_train, y_train)
echo_test = my_esn.transform(X_test)
y_true, y_pred = y_test, regr.predict(echo_test)
err = mean_squared_error(y_true, y_pred)
fp = plt.figure(figsize=(12, 4))
trainplot = fp.add_subplot(1, 3, 1)
trainplot.plot(X_train[100:600], 'b')
trainplot.set_title('Some training signal')
echoplot = fp.add_subplot(1, 3, 2)
echoplot.plot(echo_train[100:600, :20])
echoplot.set_title('Some reservoir activation')
import numpy as np
if __name__ == '__main__':
X = loadtxt('MackeyGlass_t17.txt')
X = atleast_2d(X).T
train_length = 2000
test_length = 2000
# data shape should be (lens_ts, n_features)
X_train = X[:train_length]
y_train = X[1:train_length + 1]
X_test = X[train_length:train_length + test_length]
y_test = X[train_length + 1:train_length + test_length + 1]
# Simple training
my_esn = SimpleESN(n_readout=1000,
n_components=1000,
damping=0.3,
weight_scaling=1.25)
t0 = time()
echo_train = my_esn.fit_transform(X_train)
regr = Ridge(alpha=0.01)
regr.fit(echo_train, y_train)
echo_test = my_esn.transform(X_test)
y_true, y_pred = y_test, regr.predict(echo_test)
err = mean_squared_error(y_true, y_pred)
print("done in %0.3f s" % (time() - t0))
fp = plt.figure(figsize=(12, 4))
trainplot = fp.add_subplot(1, 3, 1)
trainplot.plot(X_train[100:600], 'b')
trainplot.set_title('Some training signal')
echoplot = fp.add_subplot(1, 3, 2)
for index, i in enumerate(update):
client.send_message(f"/ESN/{index}", ((i+1)/2))
#client.send_message(f'/ESN/6', update[6])
if __name__ == '__main__':
<<<<<<< HEAD
esn = SimpleESN(n_readout=20,
n_components=20,
n_inputs=1,
input_gain=3,
input_sparcity=1,
damping=0.9,
weight_scaling=1.5,
sparcity=1.0,
random_state=31337)
=======
# esn = SimpleESN(n_readout=7, n_components=7, n_inputs=1, input_gain=6, input_sparcity=1, damping=0.9, random_state=95326, weight_scaling=1.55, sparcity=0.7)
#esn = SimpleESN(n_readout=7, n_components=7, n_inputs=1, input_gain=3, input_sparcity=1, damping=0.9, random_state=95326, weight_scaling=1.55, sparcity=0.7)
#esn = SimpleESN(n_readout=7, n_components=7, n_inputs=1, input_gain=3, input_sparcity=1, damping=0.9, random_state=95326, weight_scaling=1.25, sparcity=0.7)
#esn = SimpleESN(n_readout=25, n_components=25, n_inputs=1, input_gain=1, input_sparcity=1, damping=0.9, random_state=95326, weight_scaling=1.15, sparcity=0.95)
#esn = SimpleESN(n_readout=16, n_components=16, n_inputs=2, input_gain=3, input_sparcity=0.3, damping=0.8, random_state=95328, weight_scaling=3, sparcity=0.0685)
#esn = SimpleESN(n_readout=16, n_components=16, n_inputs=2, input_gain=3, input_sparcity=0.3, damping=0.8, random_state=95328, weight_scaling=1.15, sparcity=0.7)
#esn = SimpleESN(n_readout=16, n_components=16, n_inputs=2, input_gain=3, input_sparcity=0.3, damping=0.8, random_state=95398, weight_scaling=1.25, sparcity=0.7)
#esn = SimpleESN(n_readout=16, n_components=16, n_inputs=2, input_gain=3, input_sparcity=0.7, damping=0.9, random_state=95398, weight_scaling=1.65, sparcity=0.7)
#esn = SimpleESN(n_readout=16, n_components=16, n_inputs=2, input_gain=3, input_sparcity=0.7, damping=0.4, random_state=95398, weight_scaling=2.0, sparcity=0.7)
def test_SimpleESN_fit():
esn = SimpleESN(n_readout=n_readout)
esn = esn.fit(X)
assert_true(esn.weights_ is not None)
assert_true(esn.input_weights_ is not None)
assert_true(esn.readout_idx_ is not None)
def test_SimpleESN_incorrect_readout():
n_readout, n_components = 10, 5
esn = SimpleESN(n_readout=n_readout, n_components=n_components)
echoes = esn.fit_transform(X)
assert_true(echoes.shape == (n_samples, n_components))
def test_SimpleESN_discard():
discard_steps = 3
esn = SimpleESN(n_readout=n_readout, discard_steps=discard_steps)
echoes = esn.fit_transform(X)
assert_true(echoes.shape == (n_samples - discard_steps, n_readout))
def test_SimpleESN_shape():
esn = SimpleESN(n_readout=n_readout)
echoes = esn.fit_transform(X)
assert_true(echoes.shape == (n_samples, n_readout))
test_data = Utils.readDataFile(test_data_file) trainData=Utils.importTrainingData(training_data) testData= Utils.importTestData(test_data) trainSongsNotes = trainData[0] testSongsNotes = testData[0] trainSongsRhythm = trainData[1] testSongsRhythm = testData[1] ####################### Reservoir Part #fixed reservoir #n_readout = 3 #discard = 40 #esn = SimpleESN(n_readout, n_components = 100, damping = 0.2, weight_scaling = 0.99, random_state = 1, discard_steps = discard) esn = SimpleESN(n_readout, n_components=n_components, damping = damping, weight_scaling = weight_scaling, random_state = random_seed, discard_steps=discard) #### feed one or more songs to the reservoir and collect the echoes ####################### Learning Part possibleComposers = list(np.unique(training_data[:,1])) possibleInstruments = list(np.unique(training_data[:,3])) possibleStyles = list(np.unique(training_data[:,4])) possibleYears = list(np.unique(training_data[:,5]))
from simple_esn import SimpleESN import numpy as np n_samples, n_features = 10, 5 np.random.seed(0) X = np.random.randn(n_samples, n_features) esn = SimpleESN(n_readout=2) echoes = esn.fit_transform(X)