Pattern Attention Transformer with Doughnut Kernel

We present in this paper a new architecture, the Pattern Attention Transformer (PAT), that is composed of the new doughnut kernel. Compared with tokens in the NLP field, Transformer in computer vision has the problem of handling the high resolution of pixels in images. Inheriting the patch/window idea from ViT and its follow-ups, the doughnut kernel enhances the design of patches. It replaces the line-cut boundaries with two types of areas: sensor and updating, which is based on the comprehension of self-attention (named QKVA grid). The doughnut kernel also brings a new topic about the shape of kernels. To verify its performance on image classification, PAT is designed with Transformer blocks of regular octagon shape doughnut kernels. Its performance on ImageNet 1K surpasses the Swin Transformer (+0.7 acc1).