MLP Like

MLP-like architectures, built solely on multi-layer perceptrons, are experiencing a resurgence in computer vision and related fields, aiming to provide efficient and effective alternatives to convolutional neural networks and transformers. Current research focuses on improving their performance in challenging tasks through novel designs such as incorporating trainable product layers, multi-coordinate frame receptive fields, and frequency-aware filtering to enhance generalization and address limitations in handling spatial and temporal information. These advancements demonstrate the potential of MLP-like models for various applications, offering a compelling balance between computational efficiency and accuracy in image and video processing.

Papers

November 21, 2023

QuasiNet: a neural network with trainable product layers
Kristína Malinovská, Slavomír Holenda, Ľudovít Malinovský
Neural Network Classical Neural Network Trainable Layer Error Backpropagation MLP Like

April 8, 2023

MC-MLP:Multiple Coordinate Frames in all-MLP Architecture for Vision
Zhimin Zhu, Jianguo Zhao, Tong Mu, Yuliang Yang, Mengyu Zhu
Vision Paper Multi Layer Receptive Field MLP Architecture Single Scene Specific MLP Coordinate Frame MLP Like

March 24, 2022

FAMLP: A Frequency-Aware MLP-Like Architecture For Domain Generalization
Kecheng Zheng, Yang Cao, Kai Zhu, Ruijing Zhao, Zheng-Jun Zha
Domain Generalization Architecture Design Frequency Learning MLP Like Model Fourier Layer MLP Like

November 24, 2021

MorphMLP: An Efficient MLP-Like Backbone for Spatial-Temporal Representation Learning
David Junhao Zhang, Kunchang Li, Yali Wang, Yunpeng Chen, Shashwat Chandra, Yu Qiao, Luoqi Liu, Mike Zheng Shou
Video Representation Learning Numerous Cutting Edge Backbone MLP Like

November 7, 2021

Are we ready for a new paradigm shift? A Survey on Visual Deep MLP
Ruiyang Liu, Yinghui Li, Linmi Tao, Dun Liang, Hai-Tao Zheng
Timely Survey Computer Vision Direct Convolution Paradigm Shift MLP Like

MLP Like

Papers

QuasiNet: a neural network with trainable product layers

MC-MLP:Multiple Coordinate Frames in all-MLP Architecture for Vision

FAMLP: A Frequency-Aware MLP-Like Architecture For Domain Generalization

MorphMLP: An Efficient MLP-Like Backbone for Spatial-Temporal Representation Learning

Are we ready for a new paradigm shift? A Survey on Visual Deep MLP