def cost(X, y, beta):
'''
:params: X: N x D
:params: y: D x 1
:params: beta: weights D x 1
'''
return loss.hinge(X, y, beta)
def cost(self, X, y, beta, b):
'''
Hinge loss function
--------------------
:params: X: feature space
:params: y: target
:params: beta: weights parameters.
'''
return 0.5 * beta.dot(beta) + self.C * np.sum(loss.hinge(X, y, beta, b))
def activation(X, y, beta):
'''
:params: X: train data
:params: X: weights
'''
return 1 if loss.hinge(X, y, beta) > 0 else 0