Can Attention Enable MLPs To Catch Up With CNNs?
It addresses a foundational debate in machine learning about the sufficiency of MLPs versus CNNs, but is incremental as it reviews existing work without presenting new results.
This perspective paper examines whether recent architectures using attention mechanisms can enable multilayer perceptrons (MLPs) to match or surpass convolutional neural networks (CNNs), sparked by four 2021 studies proposing linear-based models comparable to CNNs.
In the first week of May, 2021, researchers from four different institutions: Google, Tsinghua University, Oxford University and Facebook, shared their latest work [16, 7, 12, 17] on arXiv.org almost at the same time, each proposing new learning architectures, consisting mainly of linear layers, claiming them to be comparable, or even superior to convolutional-based models. This sparked immediate discussion and debate in both academic and industrial communities as to whether MLPs are sufficient, many thinking that learning architectures are returning to MLPs. Is this true? In this perspective, we give a brief history of learning architectures, including multilayer perceptrons (MLPs), convolutional neural networks (CNNs) and transformers. We then examine what the four newly proposed architectures have in common. Finally, we give our views on challenges and directions for new learning architectures, hoping to inspire future research.