1 Hit in 1.7 sec

A Reconfigurable Winograd CNN Accelerator with Nesting Decomposition Algorithm for Computing Convolution with Large Filters [article]

Jingbo Jiang, Xizi Chen, Chi-Ying Tsui
2021 arXiv   pre-print
Recent literature found that convolutional neural networks (CNN) with large filters perform well in some applications such as image semantic segmentation.  ...  Winograd transformation helps to reduce the number of multiplications in a convolution but suffers from numerical instability when the convolution filter size gets large.  ...  it can propagate more information to the outputs in a deep neural network.  ... 
arXiv:2102.13272v1 fatcat:ubumszlwjrdalpsfy6jz2aumpe