MMD的快速计算方法

最新推荐文章于 2023-05-28 01:53:23 发布

DeniuHe

最新推荐文章于 2023-05-28 01:53:23 发布

阅读量367

点赞数

文章标签： python 机器学习人工智能

本文链接：https://blog.csdn.net/DeniuHe/article/details/130292272

版权

self._K = rbf_kernel(X=self.X, gamma=0.1)

L_len = len(self.labeled)
            U_len = len(rand_unlabeled)
            UL_matrix = self._K[np.ix_(rand_unlabeled, self.labeled)]
            UL_matrix_sum = (np.sum(UL_matrix, axis=1)+1)* (U_len-1)/(U_len + L_len)
            UU_matrix = self._K[np.ix_(rand_unlabeled, rand_unlabeled)]
            UU_matrix_sum = (np.sum(UU_matrix, axis=1)-1)* (L_len-1)/(U_len + L_len)

            MMD = np.abs(UL_matrix_sum - UU_matrix_sum)

或者

            L_len = len(self.labeled)
            U_len = len(self.unlabeled)
            UL_matrix = self._K[np.ix_(self.unlabeled, self.labeled)]
            UL_matrix_sum = (np.sum(UL_matrix, axis=1)+1)* (U_len-1)/(U_len + L_len)
            UU_matrix = self._K[np.ix_(self.unlabeled, self.unlabeled)]
            UU_matrix_sum = (np.sum(UU_matrix, axis=1)-1)* (L_len-1)/(U_len + L_len)