The kernelized version of DMLMJ.
KDMLMJ(num_dims = NULL, n_neighbors = 3, alpha = 0.001, reg_tol = 1e-10, kernel = "linear", gamma = NULL, degree = 3, coef0 = 1, kernel_params = NULL)
num_dims | Dimension desired for the transformed data. If NULL, dimension will be the number of features. |
---|---|
n_neighbors | Number of neighbors to consider in the computation of the difference spaces. |
alpha | Regularization parameter for inverse matrix computation. |
reg_tol | Tolerance threshold for applying regularization. The tolerance is compared with the matrix determinant. |
kernel | Kernel to use. Allowed values are: "linear" | "poly" | "rbf" | "sigmoid" | "cosine" | "precomputed". |
gamma | Kernel coefficient for rbf, poly and sigmoid kernels. Ignored by other kernels. Default value is 1/n_features. Float. |
degree | Degree for poly kernels. Ignored by other kernels. Integer. |
coef0 | Independent term for poly and sigmoid kernels. Ignored by other kernels. Float. |
kernel_params | Parameters (keyword arguments) and values for kernel passed as callable object. Ignored by other kernels. |
The KDMLMJ transformer, structured as a named list.
Bac Nguyen, Carlos Morell and Bernard De Baets. “Supervised distance metric learning through maximization of the Jeffrey divergence”. In: Pattern Recognition 64 (2017), pages 215-225.