SpectFormer：视觉 Transformer 中所需的频率和注意力（英文版）

英文标题：SpectFormer: Frequency and Attention is what you need in a Vision Transformer中文摘要：本研究旨在通过将谱层和多头注意力层结合起来提出 Spectformer 架构，该架构的表现优于其他转换器表示形式，特别是在图像识别任务中。英文摘要：Vision transformers have been applied successfully for image recognitiontasks. There have been either multi-headed self-attention based (ViT\cite{dosovitskiy2020image}, DeIT, \cite{touvron2021training}) similar to theoriginal work in textual models or more recently based on spectral layers(Fnet\cite{lee2021fnet}, GFNet\cite{rao2021globa