def objective(us: [float]) -> float:
model, user_search_params, info = embedding(us, n_input)
info['global_optimizer'] = global_optimizer.__name__
info['embedding_name']=embedding_name
print(' ')
print(model.summary())
search_params = {'epochs':2000,'patience':50,'jiggle_fraction':0.1,'symmetries':None}
search_params.update(user_search_params)
pprint(search_params)
model, metrics, test_error_ratio = challenge(model=model, skater_name=skater_name, k=k, n_real=60, n_samples=150, n_warm=100, n_input=n_input,
with_metrics=True, verbose=1, info=info, **search_params)
return metrics['test_error']
def build_challenger_model(n_inputs):
model = keras.Sequential()
kernel_initializer_0 = keras.initializers.RandomUniform(minval=0.01,
maxval=0.02,
seed=None)
bias_initializer_0 = keras.initializers.RandomUniform(minval=0.01,
maxval=0.21,
seed=None)
model.add(
keras.layers.Dense(80,
activation="linear",
input_shape=(1, n_inputs),
kernel_initializer=kernel_initializer_0,
bias_initializer=bias_initializer_0))
model.add(keras.layers.Dense(16, activation='linear'))
model.add(keras.layers.Dense(1, activation="linear"))
optimizer = keras.optimizers.Adagrad(learning_rate=0.005)
model.compile(loss='mse', optimizer=optimizer)
return model
if __name__ == '__main__':
skater_name = __file__.split(os.path.sep)[-1].replace('challenge_',
'').replace(
'.py', '')
print(skater_name)
model = build_challenger_model(n_inputs=80)
challenge(model=model, skater_name=skater_name, epochs=5000, patience=50)
model.add(keras.layers.Dense(6, activation="linear"))
model.add(keras.layers.Dense(8, activation="linear"))
model.add(keras.layers.Dense(1, activation="linear"))
optimizer = keras.optimizers.RMSprop(learning_rate=0.001)
model.add(keras.layers.Dropout(0.01))
model.compile(loss='mse', optimizer=optimizer)
# y = model(ones((1, n_inputs)))
return model
suggestion = """
def build_last_value_champion_model(n_inputs):
model = keras.Sequential()
model.add(keras.layers.Dense(8, activation="linear", input_shape=(1,n_inputs)))
model.add(keras.layers.Dense(6, activation="relu"))
model.add(keras.layers.Dense(8, activation="linear"))
model.add(keras.layers.Dense(1, activation="linear"))
model.compile(loss='mse')
#y = model(ones((1, n_inputs)))
return model"""
champ = """
"""
if __name__=='__main__':
skater_name = __file__.split(os.path.sep)[-1].replace('challenge_','').replace('.py','')
print(skater_name)
model = build_challenger_model(n_inputs=80)
challenge(model=model, skater_name=skater_name, epochs=50, jiggle_fraction=0.05)
model.add(keras.layers.Dense(2, activation="tanh")) # selu
model.add(keras.layers.Dense(1, activation="linear"))
optimizer = keras.optimizers.Adagrad(learning_rate=0.005)
model.compile(loss='mse', optimizer=optimizer)
return model
champ = """
def champion_model(n_inputs):
model = keras.Sequential()
kernel_initializer_0 = keras.initializers.RandomUniform(minval=0.01, maxval=0.02, seed=None)
bias_initializer_0 = keras.initializers.RandomUniform(minval=0.01, maxval=0.21, seed=None)
model.add(keras.layers.Dense(80, activation="linear", input_shape=(1, n_inputs),
kernel_initializer=kernel_initializer_0,
bias_initializer=bias_initializer_0))
model.add(keras.layers.Dense(16, activation='softsign'))
model.add(keras.layers.Dense(2, activation="tanh")) # selu
model.add(keras.layers.Dense(1, activation="linear"))
optimizer = keras.optimizers.Adagrad(learning_rate=0.005)
model.compile(loss='mse', optimizer=optimizer)
return model
"""
if __name__ == '__main__':
skater_name = __file__.split(os.path.sep)[-1].replace('challenge_',
'').replace(
'.py', '')
print(skater_name)
model = build_challenger_model(n_inputs=80)
challenge(model=model, skater_name=skater_name, epochs=500)
model.add(keras.layers.Dense(1, activation="linear"))
optimizer = keras.optimizers.RMSprop(learning_rate=0.001)
model.compile(loss='mse', optimizer=optimizer)
return model
champ = """def build_challenger_model(n_inputs):
model = keras.Sequential()
kernel_initializer_0 = keras.initializers.RandomUniform(minval=0.01, maxval=0.02, seed=None)
bias_initializer_0 = keras.initializers.RandomUniform(minval=0.01, maxval=0.21, seed=None)
model.add(keras.layers.Dense(80, activation="linear", input_shape=(1, n_inputs),
kernel_initializer=kernel_initializer_0,
bias_initializer=bias_initializer_0))
model.add(keras.layers.Dense(16, activation='linear'))
model.add(keras.layers.Dense(2, activation="linear")) # selu
model.add(keras.layers.Dense(1, activation="linear"))
optimizer = keras.optimizers.Adagrad(learning_rate=0.005)
model.compile(loss='mse', optimizer=optimizer)
return model"""
if __name__ == '__main__':
skater_name = __file__.split(os.path.sep)[-1].replace('challenge_',
'').replace(
'.py', '')
print(skater_name)
model = build_challenger_model(n_inputs=80)
challenge(model=model,
skater_name=skater_name,
epochs=5000,
patience=50,
jiggle_fraction=0.2)
from sklearned.challenging.surrogatechallenge import challenge
from tensorflow import keras
import os
def build_challenger_model(n_inputs):
model = keras.Sequential()
kernel_initializer_0 = keras.initializers.RandomUniform(minval=0.01, maxval=0.1, seed=None)
bias_initializer_0 = keras.initializers.RandomUniform(minval=-0.2, maxval=0.2, seed=None)
model.add(keras.layers.Dense(80, activation="linear", input_shape=(1, n_inputs),
kernel_initializer=kernel_initializer_0,
bias_initializer=bias_initializer_0))
model.add(keras.layers.Dense(16, activation='linear'))
model.add(keras.layers.Dense(2, activation="linear")) # selu
model.add(keras.layers.Dense(1, activation="linear"))
optimizer = keras.optimizers.RMSprop(learning_rate=0.001)
model.compile(loss='mse', optimizer=optimizer)
return model
if __name__=='__main__':
skater_name = __file__.split(os.path.sep)[-1].replace('challenge_','').replace('.py','')
print(skater_name)
model = build_challenger_model(n_inputs=80)
challenge(model=model, skater_name=skater_name, epochs=5, patience=25)