ailayer_conv2d_default.h File Reference

Default implementation of the Conv2D layer . More...

Go to the source code of this file.

Typedefs
typedef struct ailayer_conv2d	ailayer_conv2d_f32_t

Functions
ailayer_t *	ailayer_conv2d_f32_default (ailayer_conv2d_f32_t *layer, ailayer_t *input_layer)
	Initializes and connect a Conv2D layer with the F32 default implementation. More...
ailayer_t *	ailayer_conv2d_cfirst_f32_default (ailayer_conv2d_f32_t *layer, ailayer_t *input_layer)
	Initializes and connect a Conv2D layer (channels first) with the F32 default implementation. More...
ailayer_t *	ailayer_conv2d_chw_f32_default (ailayer_conv2d_f32_t *layer, ailayer_t *input_layer)
	Initializes and connect a Conv2D layer (channels first) with the F32 default implementation. More...
ailayer_t *	ailayer_conv2d_clast_f32_default (ailayer_conv2d_f32_t *layer, ailayer_t *input_layer)
	Initializes and connect a Conv2D layer (channels last) with the F32 default implementation. More...
ailayer_t *	ailayer_conv2d_hwc_f32_default (ailayer_conv2d_f32_t *layer, ailayer_t *input_layer)
	Initializes and connect a Conv2D layer (channels last) with the F32 default implementation. More...
void	ailayer_conv2d_init_params_f32_default (ailayer_t *self)
	F32 default implementation of the ailayer.init_params function for the Conv2D layer More...

Detailed Description

Default implementation of the Conv2D layer .

Version

2.2.0

Date

18.10.2021

Copyright

Copyright (C) 2020-2023 Fraunhofer Institute for Microelectronic Circuits and Systems. All rights reserved.

AIfES is free software: you can redistribute it and/or modify it under the terms of the GNU Affero General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Affero General Public License for more details.

You should have received a copy of the GNU Affero General Public License along with this program. If not, see https://www.gnu.org/licenses/.

Hardware independent implementations of the Conv2D layer in F32 data-type. For more information about the Conv2D layer refer to ailayer_conv2d.h.

Function Documentation

ailayer_conv2d_cfirst_f32_default()

ailayer_t* ailayer_conv2d_cfirst_f32_default	(	ailayer_conv2d_f32_t *	layer,
		ailayer_t *	input_layer
	)

Initializes and connect a Conv2D layer (channels first) with the F32 default implementation.

Code examples are given in the description of ailayer_conv2d_f32_default().

Parameters

*layer	The layer structure to initialize.
*input_layer	The prior layer.

Returns

The (successfully) initialized layer structure.

ailayer_conv2d_chw_f32_default()

ailayer_t* ailayer_conv2d_chw_f32_default	(	ailayer_conv2d_f32_t *	layer,
		ailayer_t *	input_layer
	)

Initializes and connect a Conv2D layer (channels first) with the F32 default implementation.

Code examples are given in the description of ailayer_conv2d_f32_default().

Parameters

*layer	The layer structure to initialize.
*input_layer	The prior layer.

Returns

The (successfully) initialized layer structure.

ailayer_conv2d_clast_f32_default()

ailayer_t* ailayer_conv2d_clast_f32_default	(	ailayer_conv2d_f32_t *	layer,
		ailayer_t *	input_layer
	)

Initializes and connect a Conv2D layer (channels last) with the F32 default implementation.

Code examples are given in the description of ailayer_conv2d_f32_default().

Parameters

*layer	The layer structure to initialize.
*input_layer	The prior layer.

Returns

The (successfully) initialized layer structure.

ailayer_conv2d_f32_default()

ailayer_t* ailayer_conv2d_f32_default	(	ailayer_conv2d_f32_t *	layer,
		ailayer_t *	input_layer
	)

Initializes and connect a Conv2D layer with the F32 default implementation.

Example: Create the layer structure with pretrained weights:
In C:

// Use constant data only for inference. For training remove the const qualifier!!
const float kernel_data_conv2d[] = {0.0f, -0.1f,
                                    0.1f, 0.2f,
 
                                    -0.2f, 0.2f,
                                    -0.1f, 0.0f};
const float bias_data_conv2d[] = {0.0f, 0.0f};
ailayer_conv2d_f32_t conv2d_layer = {
    .filter_count =     3,
    .kernel_size =      {2, 2},
    .stride =           {1, 1},
    .dilation =         {1, 1},
    .padding =          {0, 0},
    .weights.data =     kernel_data_conv2d,
    .bias.data =        bias_data_conv2d
};

In C, C++ and on Arduino:

// Use constant data only for inference. For training remove the const qualifier!!
const float kernel_data_conv2d[] = {0.0f, -0.1f,
                                    0.1f, 0.2f,
 
                                    -0.2f, 0.2f,
                                    -0.1f, 0.0f};
const float bias_data_conv2d[] = {0.0f, 0.0f};
ailayer_conv2d_t conv2d_layer         = AILAYER_CONV2D_F32_M(
                                                                /* filters =*/     2,
                                                                /* kernel_size =*/ HW(2, 2),
                                                                /* stride =*/      HW(1, 1),
                                                                /* dilation =*/    HW(1, 1),
                                                                /* padding =*/     HW(0, 0),
                                                                /* weights =*/     kernel_data_conv2d,
                                                                /* bias =*/        bias_data_conv2d
                                                             );

Example: Create the layer structure for automatic parameter distribution:
In C:

ailayer_conv2d_f32_t conv2d_layer = {
    .filter_count =     3,
    .kernel_size =      {2, 2},
    .stride =           {1, 1},
    .dilation =         {1, 1},
    .padding =          {0, 0}
};

In C, C++ and on Arduino:

ailayer_conv2d_t conv2d_layer         = AILAYER_CONV2D_F32_A(
                                                                /* filters =*/     2,
                                                                /* kernel_size =*/ HW(2, 2),
                                                                /* stride =*/      HW(1, 1),
                                                                /* dilation =*/    HW(1, 1),
                                                                /* padding =*/     HW(0, 0)
                                                             );

Example: Initialize and connect the layer for data with channels first (all options are equivalent):

conv2d_layer.channel_axis = 1;
x = ailayer_conv2d_f32_default(&conv2d_layer, x);

or

x = ailayer_conv2d_cfirst_f32_default(&conv2d_layer, x);

or

x = ailayer_conv2d_chw_f32_default(&conv2d_layer, x);

Example: Initialize and connect the layer for data with channels last (all options are equivalent):

conv2d_layer.channel_axis = -1;
x = ailayer_conv2d_f32_default(&conv2d_layer, x);

or

x = ailayer_conv2d_clast_f32_default(&conv2d_layer, x);

or

x = ailayer_conv2d_hwc_f32_default(&conv2d_layer, x);

Parameters

*layer	The layer structure to initialize.
*input_layer	The prior layer.

Returns

The (successfully) initialized layer structure.

ailayer_conv2d_hwc_f32_default()

ailayer_t* ailayer_conv2d_hwc_f32_default	(	ailayer_conv2d_f32_t *	layer,
		ailayer_t *	input_layer
	)

Initializes and connect a Conv2D layer (channels last) with the F32 default implementation.

Code examples are given in the description of ailayer_conv2d_f32_default().

Parameters

*layer	The layer structure to initialize.
*input_layer	The prior layer.

Returns

The (successfully) initialized layer structure.

ailayer_conv2d_init_params_f32_default()

void ailayer_conv2d_init_params_f32_default

(

ailayer_t *

self

)

F32 default implementation of the ailayer.init_params function for the Conv2D layer

Implementation of ailayer.init_params.

The function will initialize the weights and bias depending on the following activation function.

Activation function	Weights-init	Bias-init
None, tanh, logistic, softmax	Glorot	Zeros
ReLu and variants	He	Zeros

Parameters

*self

The layer structure