def ml_problem(self):
np.random.seed(0)
X = np.random.standard_normal((self.num_rows, self.num_cols))
w = self.random_weights()
y = np.argmax(_group_lasso._softmax(X @ w), axis=1)
y = LabelBinarizer().fit_transform(y)
return X, y, w
def sparse_ml_problem(self):
X = sparse.random(self.num_rows, self.num_cols, random_state=0)
X = sparse.dok_matrix(X)
for row in range(self.num_rows):
col = np.random.randint(self.num_cols)
X[row, col] = np.random.standard_normal()
w = self.random_weights()
y = np.argmax(_group_lasso._softmax(X @ w), axis=1)
y = LabelBinarizer().fit_transform(y)
return X, y, w