Towards a General Purpose CNN for Long Range Dependencies in ND

论文地址：

• https://arxiv.org/pdf/2206.03398.pdf

整体思路以及计算方式

将CKConv的方式推广到了高维，区别在于将映射修改为$\mathbb Z^n \to \mathbb R$。

代码

• https://github.com/david-knigge/ccnn