C++ API Reference for Intel® Data Analytics Acceleration Library 2020 Update 1
Parameters for the forward and backward batch normalization layers. More...
Parameter | ( | double | alpha = 0.01 , |
double | epsilon = 0.00001 , |
||
size_t | dimension = 1 |
||
) |
Constructs the parameters of the batch normalization layer
[in] | alpha | Smoothing factor that is used in population mean and population variance computations |
[in] | epsilon | A constant added to the mini-batch variance for numerical stability |
[in] | dimension | Index of the dimension for which the normalization is performed |
double alpha |
Smoothing factor that is used in population mean and population variance computations
size_t dimension |
Index of the dimension for which the normalization is performed
double epsilon |
A constant added to the mini-batch variance for numerical stability
For more complete information about compiler optimizations, see our Optimization Notice.