CMSIS-NN  Version 3.0.0
CMSIS NN Software Library
 All Data Structures Files Functions Variables Enumerations Enumerator Macros Groups Pages
Activation Functions

Functions

void arm_nn_activations_direct_q15 (q15_t *data, uint16_t size, uint16_t int_width, arm_nn_activation_type type)
 neural network activation function using direct table look-up More...
 
void arm_nn_activations_direct_q7 (q7_t *data, uint16_t size, uint16_t int_width, arm_nn_activation_type type)
 Q7 neural network activation function using direct table look-up. More...
 
void arm_relu6_s8 (q7_t *data, uint16_t size)
 s8 ReLU6 function More...
 
void arm_relu_q15 (q15_t *data, uint16_t size)
 Q15 RELU function. More...
 
void arm_relu_q7 (q7_t *data, uint16_t size)
 Q7 RELU function. More...
 

Description

Perform activation layers, including ReLU (Rectified Linear Unit), sigmoid and tanh

Function Documentation

void arm_nn_activations_direct_q15 ( q15_t *  data,
uint16_t  size,
uint16_t  int_width,
arm_nn_activation_type  type 
)

Q15 neural network activation function using direct table look-up.

Note
Refer header file for details.

References ARM_SIGMOID, ARM_TANH, sigmoidTable_q15, and tanhTable_q15.

void arm_nn_activations_direct_q7 ( q7_t *  data,
uint16_t  size,
uint16_t  int_width,
arm_nn_activation_type  type 
)
Parameters
[in,out]datapointer to input
[in]sizenumber of elements
[in]int_widthbit-width of the integer part, assume to be smaller than 3
[in]typetype of activation functions

This is the direct table look-up approach.

Assume here the integer part of the fixed-point is <= 3. More than 3 just not making much sense, makes no difference with saturation followed by any of these activation functions.

References ARM_SIGMOID, ARM_TANH, sigmoidTable_q7, and tanhTable_q7.

void arm_relu6_s8 ( q7_t *  data,
uint16_t  size 
)
Parameters
[in,out]datapointer to input
[in]sizenumber of elements

References MAX, and MIN.

void arm_relu_q15 ( q15_t *  data,
uint16_t  size 
)
Parameters
[in,out]datapointer to input
[in]sizenumber of elements

Optimized relu with QSUB instructions.

References arm_nn_write_q15x2_ia().

void arm_relu_q7 ( q7_t *  data,
uint16_t  size 
)
Parameters
[in,out]datapointer to input
[in]sizenumber of elements

Optimized relu with QSUB instructions.