no code implementations • 26 Jan 2023 • Ziran Qin, Mingbao Lin, Weiyao Lin
This paper focuses on Winograd transformation in 3D convolutional neural networks (CNNs) that are more over-parameterized compared with the 2D version.