nntk_ffd_eval_func
Evaluate the output of a defined feed-forward neural network.
Utility Routine Interface
Fortran usage:
include 'aba_nntk.hdr'
integer iRetCode
...
iRetCode = nntk_ffd_eval_func ( nnet_id, x_vals, y_vals )
or C/C++ usage:
#include <aba_nntk.h>
int nntk_ffd_eval_funcC(const int& nnet_id, const double* x_vals, double* y_vals);
Variables to Be Provided to the Utility Routine
-
nnet_id
-
The integer identifier of the neural network to be evaluated.
-
x_vals
-
Array containing the input values
at which the neural network will be evaluated. The size of this
input array is assumed to be the same as the neural network input size
specified in the input file.
Variables Returned from the Utility Routine
-
y_vals
-
Array in which the neural network output values
will be stored. The size of this output array is assumed to be the
same as the neural network output size
specified in the input file.
-
iRetCode
-
Integer return code value. The evaluation utility returns 0 if no errors are
detected; otherwise, it returns a nonzero value.
nntk_ffd_eval_deriv
Evaluate the Jacobian matrix (derivatives of the outputs with respect to inputs) for a
defined feed-forward neural network.
Utility Routine Interface
Fortran usage:
include 'aba_nntk.hdr'
integer iRetCode
...
iRetCode = nntk_ffd_eval_deriv ( nnet_id, x_vals, jac_vals )
or C/C++ usage:
#include <aba_nntk.h>
int nntk_ffd_eval_derivC(const int& nnet_id, const double* x_vals, double* jac_vals);
Variables to Be Provided to the Utility Routine
-
nnet_id
-
The integer identifier of the neural network to be evaluated.
-
x_vals
-
Array containing the input values
at which the Jacobian matrix of the neural network will be
evaluated. The size of this input array is assumed to be the same as the neural
network input size
specified in the input file.
Variables Returned from the Utility Routine
-
jac_vals
-
Array in which the Jacobian matrix entries will be stored. The size of this output
array should be equal to the product of the neural network output size
and the neural network input size
. The Jacobian matrix entries are written in the following order:
-
iRetCode
-
Integer valued return code. The evaluation utility returns 0 if no errors are
detected; otherwise, it returns a nonzero value.
nntk_ffd_eval_func_deriv
Evaluate the output and the Jacobian matrix (derivatives of the outputs with respect to
inputs) for a defined feed-forward neural network.
Utility Routine Interface
Fortran usage:
include 'aba_nntk.hdr'
integer iRetCode
...
iRetCode = nntk_ffd_eval_func_deriv ( nnet_id, x_vals, y_vals, jac_vals )
or C/C++ usage:
#include <aba_nntk.h>
int nntk_ffd_eval_func_derivC(const int& nnet_id, const double* x_vals, double* y_vals, double* jac_vals);
Variables to Be Provided to the Utility Routine
-
nnet_id
-
The integer identifier of the neural network to be evaluated.
-
x_vals
-
Array containing the input values
at which the output and the Jacobian matrix of the neural network
will be evaluated. The size of this input array is assumed to be the same as the
neural network input size
specified in the input file.
Variables Returned from the Utility Routine
-
y_vals
-
Array in which the neural network output values
will be stored. The size of this output array is assumed to be the
same as the neural network output size
specified in the input file.
-
jac_vals
-
Array in which the Jacobian matrix entries will be stored. The size of this output
array should be equal to the product of the neural network output size
and the neural network input size
. The Jacobian matrix entries are written in the following order:
-
iRetCode
-
Integer valued return code. The evaluation utility returns 0 if no errors were
detected; otherwise, it returns a nonzero value.
|