def fit(self, A, y, group_ids):
x = np.zeros(A.shape[1])
theta = x.copy()
gamma = x.copy()
inv_matrix = np.linalg.inv(A.T @ A + self.rho)
group_ids = LabelBinarizer().fit_transform(group_ids)
group_ids = group_ids.astype(np.float64)
for _ in range(self.max_iter):
x = self.update_x(A, y, gamma, theta, inv_matrix)
theta_hat = self.update_theta_hat(x, gamma)
w = self.update_w(theta_hat)
theta = self.update_theta(w, theta_hat, group_ids)
gamma = self.update_gamma(A, x, gamma, theta)
self.coef_ = theta
return self
def fit(self, A, y, group_ids):
# 標準化
if self.normalize == True:
sscaler = StandardScaler()
sscaler.fit(A)
A = sscaler.transform(A)
# 初期値、ハイパラ
n_samples = A.shape[0]
w = np.zeros(A.shape[1])
rho = self.supermum_eigen(A.T @ A)
alpha = self.alpha * n_samples
# group-ids -> one-hot vector
group_ids = LabelBinarizer().fit_transform(group_ids)
group_ids = group_ids.astype(np.float64)
group_id_sum = np.sum(group_ids, axis=0)
for it in range(self.max_iter):
w = self.update(w, A, y, rho, alpha, group_ids, group_id_sum)
self.coef = w
return self
def one_hot_encode_labels(data_y):
data_y = LabelBinarizer().fit_transform(data_y)
data_y = data_y.astype('float32')
return data_y
train_x = train_x.reshape(-1, 3, 32, 32)
train_y = train_y.reshape(-1, 1)
test_x = test_x.reshape(-1, 3, 32, 32)
test_y = test_y.reshape(-1, 1)
train_x = train_x / 255.
test_x = test_x / 255.
# Binalize
labels_train = LabelBinarizer().fit_transform(train_y)
labels_test = LabelBinarizer().fit_transform(test_y)
# Change types.
train_x = train_x.astype(np.float32)
test_x = test_x.astype(np.float32)
labels_train = labels_train.astype(np.float32)
labels_test = labels_test.astype(np.float32)
N = len(train_x)
class Cifar10(rm.Model):
def __init__(self):
super(Cifar10, self).__init__()
self._l1 = rm.Conv2d(channel=32)
self._l2 = rm.Conv2d(channel=32)
self._l3 = rm.Conv2d(channel=64)
self._l4 = rm.Conv2d(channel=64)
self._l5 = rm.Dense(512)
self._l6 = rm.Dense(10)
self._sd = rm.SpatialDropout(dropout_ratio=0.25)
