SuperNN  1.0.0
Public Attributes | List of all members
SuperNN::NBN::Hist Struct Reference

Public Attributes

Eigen::VectorXd gradient
 Error gradient matrix. More...
 
Eigen::MatrixXd hessian
 Quasi-Hessian matrix. More...
 
Eigen::VectorXd t_jacob_line
 Transposed Jacobian line. More...
 
Eigen::VectorXd weights
 Weight matrix (same as the network weights, but in a suitable format to the algorithm. More...
 
Eigen::VectorXd weights_backup
 Backup of the weights, for backtracking. More...
 

Detailed Description

Definition at line 385 of file training.cpp.

Member Data Documentation

Eigen::VectorXd SuperNN::NBN::Hist::gradient

Error gradient matrix.

Definition at line 394 of file training.cpp.

Eigen::MatrixXd SuperNN::NBN::Hist::hessian

Quasi-Hessian matrix.

Definition at line 388 of file training.cpp.

Eigen::VectorXd SuperNN::NBN::Hist::t_jacob_line

Transposed Jacobian line.

Definition at line 391 of file training.cpp.

Eigen::VectorXd SuperNN::NBN::Hist::weights

Weight matrix (same as the network weights, but in a suitable format to the algorithm.

Definition at line 398 of file training.cpp.

Eigen::VectorXd SuperNN::NBN::Hist::weights_backup

Backup of the weights, for backtracking.

Definition at line 401 of file training.cpp.