Intel(R) Math Kernel Library for Deep Neural Networks (Intel(R) MKL-DNN)
0.13
Performance library for Deep Learning
|
A primitive to perform batch normalization
. More...
Functions | |
mkldnn_status_t MKLDNN_API | mkldnn_batch_normalization_forward_desc_init (mkldnn_batch_normalization_desc_t *bnrm_desc, mkldnn_prop_kind_t prop_kind, const mkldnn_memory_desc_t *data_desc, float epsilon, unsigned flags) |
Initializes a batch normalization descriptor bnrm_desc for forward propagation using prop_kind , (possible values are mkldnn_forward_training or mkldnn_forward_inference), memory descriptor data_desc , normalization parameter epsilon and flags (possible values are mkldnn_use_global_stats and mkldnn_use_scaleshift). More... | |
mkldnn_status_t MKLDNN_API | mkldnn_batch_normalization_backward_desc_init (mkldnn_batch_normalization_desc_t *bnrm_desc, mkldnn_prop_kind_t prop_kind, const mkldnn_memory_desc_t *diff_data_desc, const mkldnn_memory_desc_t *data_desc, float epsilon, unsigned flags) |
Initializes a batch normalization descriptor bnrm_desc for backward propagation with respect to data and scale-shift parameters using memory descriptors data_desc and diff_data_desc , and normalization parameter epsilon and flags (possible values are mkldnn_use_global_stats and mkldnn_use_scaleshift). More... | |
A primitive to perform batch normalization
.
where are weights and bias for a channel and,
,
,
and eps is a constant to improve numerical stability.
mkldnn_status_t MKLDNN_API mkldnn_batch_normalization_backward_desc_init | ( | mkldnn_batch_normalization_desc_t * | bnrm_desc, |
mkldnn_prop_kind_t | prop_kind, | ||
const mkldnn_memory_desc_t * | diff_data_desc, | ||
const mkldnn_memory_desc_t * | data_desc, | ||
float | epsilon, | ||
unsigned | flags | ||
) |
Initializes a batch normalization descriptor bnrm_desc
for backward propagation with respect to data and scale-shift parameters using memory descriptors data_desc
and diff_data_desc
, and normalization parameter epsilon
and flags (possible values are mkldnn_use_global_stats and mkldnn_use_scaleshift).
mkldnn_status_t MKLDNN_API mkldnn_batch_normalization_forward_desc_init | ( | mkldnn_batch_normalization_desc_t * | bnrm_desc, |
mkldnn_prop_kind_t | prop_kind, | ||
const mkldnn_memory_desc_t * | data_desc, | ||
float | epsilon, | ||
unsigned | flags | ||
) |
Initializes a batch normalization descriptor bnrm_desc
for forward propagation using prop_kind
, (possible values are mkldnn_forward_training or mkldnn_forward_inference), memory descriptor data_desc
, normalization parameter epsilon
and flags (possible values are mkldnn_use_global_stats and mkldnn_use_scaleshift).