laplacian_svm.py 文件源码

def fit(self, X, y, L):
        """Fit the model according to the given training data.

        Prameters
        ---------
        X : array-like, shpae = [n_samples, n_features]
            Training data.

        y : array-like, shpae = [n_samples]
            Target values (unlabeled points are marked as 0).

        L : array-like, shpae = [n_samples, n_samples]
            Graph Laplacian.
        """

        labeled               = y != 0
        y_labeled             = y[labeled]
        n_samples, n_features = X.shape
        n_labeled_samples     = y_labeled.size
        I                     = sp.eye(n_samples)
        Y                     = sp.diags(y_labeled)
        J                     = sp.eye(n_labeled_samples, n_samples)
        K                     = rbf_kernel(X, gamma=self.gamma_k)
        M                     = 2 * self.gamma_a * I \
            + 2 * self.gamma_i / n_samples**2 * L**self.p @ K

        # Construct the QP, invoke solver
        solvers.options['show_progress'] = False
        sol                   = solvers.qp(
            P                 = matrix(Y @ J @ K @ LA.inv(M) @ J.T @ Y),
            q                 = matrix(-1 * np.ones(n_labeled_samples)),
            G                 = matrix(np.vstack((
                -1 * np.eye(n_labeled_samples),
                n_labeled_samples * np.eye(n_labeled_samples)
            ))),
            h                 = matrix(np.hstack((
                np.zeros(n_labeled_samples),
                np.ones(n_labeled_samples)
            ))),
            A                 = matrix(y_labeled, (1, n_labeled_samples), 'd'),
            b                 = matrix(0.0)
        )

        # Train a classifer
        self.dual_coef_       = LA.solve(M, J.T @ Y @ np.array(sol['x']).ravel())

        return self