Abstract: Recently, MLP-based architectures have achieved competitive performance with convolutional neural networks (CNNs) and vision transformers (ViTs) across various vision tasks. However, most ...