![]() |
Neuron Net
|
Describes the gradient calculation process for the Neuron of convolution layer. More...
Macros | |
#define | def_k_CalcHiddenGradientConv 8 |
Index of the kernel of the convolution neuron to transfer gradient to previous layer (CalcHiddenGradientConv) More... | |
#define | def_k_chgc_matrix_w 0 |
Weights matrix (m+1)*n, where m - input window and n - output window. More... | |
#define | def_k_chgc_matrix_g 1 |
Tensor of gradients at current layer. More... | |
#define | def_k_chgc_matrix_o 2 |
Output tensor. More... | |
#define | def_k_chgc_matrix_ig 3 |
Tensor of gradients at previous layer. More... | |
#define | def_k_chgc_outputs 4 |
Number of outputs. More... | |
#define | def_k_chgc_step 5 |
Step size. More... | |
#define | def_k_chgc_window_in 6 |
Size of input window. More... | |
#define | def_k_chgc_window_out 7 |
Size of output window. More... | |
#define | def_k_chgc_activation 8 |
Activation type (ENUM_ACTIVATION) More... | |
Functions | |
__kernel void | CalcHiddenGradientConv (__global double *matrix_w, __global double *matrix_g, __global double *matrix_o, __global double *matrix_ig, int outputs, int step, int window_in, int window_out, uint activation) |
Kernel of the Convolution neuron to transfer gradient to previous layer (CNeuronConvOCL) More... | |
Describes the gradient calculation process for the Neuron of convolution layer.
#define def_k_CalcHiddenGradientConv 8 |
Index of the kernel of the convolution neuron to transfer gradient to previous layer (CalcHiddenGradientConv)
Definition at line 190 of file NeuroNet.mqh.
#define def_k_chgc_activation 8 |
Activation type (ENUM_ACTIVATION)
Definition at line 199 of file NeuroNet.mqh.
#define def_k_chgc_matrix_g 1 |
Tensor of gradients at current layer.
Definition at line 192 of file NeuroNet.mqh.
#define def_k_chgc_matrix_ig 3 |
Tensor of gradients at previous layer.
Definition at line 194 of file NeuroNet.mqh.
#define def_k_chgc_matrix_o 2 |
Output tensor.
Definition at line 193 of file NeuroNet.mqh.
#define def_k_chgc_matrix_w 0 |
Weights matrix (m+1)*n, where m - input window and n - output window.
Definition at line 191 of file NeuroNet.mqh.
#define def_k_chgc_outputs 4 |
Number of outputs.
Definition at line 195 of file NeuroNet.mqh.
#define def_k_chgc_step 5 |
Step size.
Definition at line 196 of file NeuroNet.mqh.
#define def_k_chgc_window_in 6 |
Size of input window.
Definition at line 197 of file NeuroNet.mqh.
#define def_k_chgc_window_out 7 |
Size of output window.
Definition at line 198 of file NeuroNet.mqh.
__kernel void CalcHiddenGradientConv | ( | __global double * | matrix_w, |
__global double * | matrix_g, | ||
__global double * | matrix_o, | ||
__global double * | matrix_ig, | ||
int | outputs, | ||
int | step, | ||
int | window_in, | ||
int | window_out, | ||
uint | activation | ||
) |
Kernel of the Convolution neuron to transfer gradient to previous layer (CNeuronConvOCL)
[in] | matrix_w | Weights matrix (m+1)*n, where m - input window and n - output window |
[in] | matrix_g | Tensor of gradients at current layer |
[in] | matrix_o | Output tensor |
[out] | matrix_ig | Tensor of gradients at previous layer |
outputs | Number of outputs | |
step | Step size | |
window_in | Size of input window | |
window_out | Size of output window | |
activation | Activation type (ENUM_ACTIVATION) |
Definition at line 395 of file NeuroNet.cl.