2021/12/07 by Haofei Zhang, Zhang, Haofei, Jiarui Duan +9
Computer Science · Engineering · Neuroscience · #Advanced Neural Network Applications #Artificial Intelligence (cs.AI) #Brain Tumor Detection and Classification #CCD and CMOS Imaging Sensors #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #cs.AI #cs.CV
paper · pdf · doi:10.48550/arxiv.2112.03552
Accepted as a conference paper by CVPR2022
openalex publication_date 2021/12/07 · openalex created_date 2021/12/31 · arxiv created 2022/03/26 · arxiv updated 2022/03/29 · openalex updated_date 2026/07/28
Recently, vision Transformers (ViTs) are developing rapidly and starting to challenge the domination of convolutional neural networks (CNNs) in the realm of computer vision (CV). With the general-purpose Transformer architecture replacing the hard-coded inductive biases of convolution, ViTs have surpassed CNNs, especially in data-sufficient circumstances. However, ViTs are prone to over-fit on small datasets and thus rely on large-scale pre-training, which expends enormous time. In this paper, we strive to liberate ViTs from pre-training by introducing CNNs' inductive biases back to ViTs while preserving their network architectures for higher upper bound and setting up more suitable optimization objectives. To begin with, an agent CNN is designed based on the given ViT with inductive biases. Then a bootstrapping training algorithm is proposed to jointly optimize the agent and ViT with weight sharing, during which the ViT learns inductive biases from the intermediate features of the agent. Extensive experiments on CIFAR-10/100 and ImageNet-1k with limited training data have shown encouraging results that the inductive biases help ViTs converge significantly faster and outperform conventional CNNs with even fewer parameters. Our code is publicly available at https://github.com/zhfeing/Bootstrapping-ViTs-pytorch.