dbl_val: 0.05
name: "dropout5"
dbl_val: 0.05
name: "dropout6"
dbl_val: 0.05
name: "regularizer1"
dbl_val: 0.0001
name: "regularizer2"
dbl_val: 0.0001
name: "regularizer3"
dbl_val: 0.0001
name: "regularizer4"
dbl_val: 0.0001
"""
select_gpu()
params = [
10, 10, 10, 10, 10, 10, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.0001, 0.0001,
0.0001, 0.0001
]
x_train, y_train, x_test, y_test = get_data()
X = np.concatenate((x_train, x_test))
Y = np.concatenate((y_train, y_test))
k_fold = StratifiedKFold(n_splits=10, shuffle=False, random_state=7)
ret = []
y_stub = np.random.randint(0, 10, X.shape[0])
for train, test in k_fold.split(X, y_stub):
ret.append(cnn(params, (X[train], Y[train], X[test], Y[test])))
print(np.array(ret))
def run(params):
print('Start time: ', datetime.datetime.now())
result = cnn(params, get_data())
print('Result: ', params, result)
print('End time: ', datetime.datetime.now())
return result
from examples.cnn.mnist.mnist import get_data
from examples.cnn.cnn import cnn, select_gpu
select_gpu()
for width in [64, 128, 256]:
for dropout in [0.25, 0.5]:
for regularizer in [0.01, 0.001]:
params = [width * 1.0] * 6 + [dropout] * 6 + [regularizer] * 4
print(params)
cnn(params, get_data())
def run(params):
print('Start time: ', datetime.datetime.now())
result = cnn(params, get_data(), Conv1D, MaxPooling1D, True, True)
print('Result: ', params, result)
print('End time: ', datetime.datetime.now())
return result
