FPGA architecture for convolutional neural network training
- 10 October 2023
- conference paper
- Published by SPIE-Intl Soc Optical Eng
- Vol. 12799, 952-957
Abstract
Convolutional neural networks (CNNs) have been gradually applied to aerospace equipment in recent years. However, CNN requires a significant amount of hardware resources, while parameters need to be updated for different tasks in space. We propose an FPGA architecture for CNN training through a multi-module collaborative design strategy. On the basis of the intra-layer parallelism, we analyze the data storage and the operations to deduce forward and backward propagation on embedded devices. Based on the experimental results, the working frequency can reach 250 MHz and the performance is 15.9 GOPS. The resource usage and efficiency of this design are superior to those of other similar hardware acceleration platforms.Keywords
This publication has 10 references indexed in Scilit:
- EF-Train: Enable Efficient On-device CNN Training on FPGA through Data Reshaping for Online Adaptation or PersonalizationACM Transactions on Design Automation of Electronic Systems, 2022
- ETA: An Efficient Training Accelerator for DNNs Based on Hardware-Algorithm Co-OptimizationIEEE Transactions on Neural Networks and Learning Systems, 2022
- Cambricon-Q: A Hybrid Architecture for Efficient Training2021
- Deep Neural Network Training Accelerator Designs in ASIC and FPGA2020
- Challenges in Energy-Efficient Deep Neural Network Training with FPGA2020
- Towards efficient deep neural network training by FPGA-based batch-level parallelismJournal of Semiconductors, 2020
- An FPGA-based processor for training convolutional neural networks2017
- F-CNN: An FPGA-based framework for training Convolutional Neural Networks2016
- CNN-MERP: An FPGA-based memory-efficient reconfigurable processor for forward and backward propagation of convolutional neural networks2016
- Gradient-based learning applied to document recognitionProceedings of the IEEE, 1998