Instituto Tecnológico de Celaya Tecnológico Nacional de México
ITC Investigación ITC
Explorar publicaciones
Artículo · 2022 · Catálogo consolidado

Flexible Convolver for Convolutional Neural Networks Deployment onto Hardware-Oriented Applications

Applied Sciences

Moisés Arredondo-Velázquez ORCID Faculty of Physical and Mathematical Sciences, Benemérita Universidad Autónoma de Puebla, Puebla 72410, Mexico
Paulo Aaron Aguirre-Álvarez ORCID Electronics Engineering Department, Tecnológico Nacional de México en Celaya, Celaya 38010, Mexico
Alfredo Padilla-Medina ORCID Electronics Engineering Department, Tecnológico Nacional de México en Celaya, Celaya 38010, Mexico
Alejandro Espinosa-Calderon ORCID Regional Center for Optimization and Development of Equipment, Tecnológico Nacional de México, Celaya 38020, Mexico
Juan Prado-Olivarez ORCID Electronics Engineering Department, Tecnológico Nacional de México en Celaya, Celaya 38010, Mexico
Javier Diaz-Carmona ORCID Electronics Engineering Department, Tecnológico Nacional de México en Celaya, Celaya 38010, Mexico

Resumen

This paper introduces a flexible convolver capable of adapting to the different convolution layer configurations of state-of-the-art Convolution Neural Networks (CNNs). The use of two proposed programmable components achieves this adaptability. A Programmable Line Buffer (PLB) based on Programmable Shift Registers (PSRs) allows the generation of the required convolution masks required for each processed CNN layer. The convolution layer computing is performed through a proposed programmable systolic array configured according to the target device resources. In order to maximize the device resource usage and to achieve a shortened processing time, the filter, data, and loop parallelisms are leveraged. These characteristics allow the described architecture to be scalable and implemented on any FPGA device targeting different applications. The convolver description was written in VHDL using the Intel Cyclone V 5CSXFC6D6F31C6N device as a reference. The experimental results show that the proposed computing method allows the processing of any CNN without requiring special adaptation for a specific application since the standard convolution algorithm is used. The proposed flexible convolver achieves competitive performance compared with those reported in related works.

Datos de publicación

Cita esta publicación

Comprueba el estilo que requiere tu institución o revista.

Moisés Arredondo-Velázquez; Paulo Aaron Aguirre-Álvarez; Alfredo Padilla-Medina; Alejandro Espinosa-Calderon; Juan Prado-Olivarez; Javier Diaz-Carmona. (2022). Flexible Convolver for Convolutional Neural Networks Deployment onto Hardware-Oriented Applications. Applied Sciences, 13(1), 93. https://doi.org/10.3390/app13010093

Fuentes y trazabilidad

Se conserva la procedencia de los datos y se da prioridad a la revista cuando existe información editorial.

Licencias declaradas por las fuentes

https://creativecommons.org/licenses/by/4.0/

Descargar ficha bibliográfica

El PDF contiene los metadatos y el enlace al trabajo original.