Global routing between capsules

Ran Chen

doi:10.1016/j.patcog.2023.110142

2024-04-01

PRE

DOI：10.1016/j.patcog.2023.110142

原文链接

原文求助

摘要

En 中文

Current convolutional neural networks (CNNs) lack viewpoint equivariance. Hence, they perform poorly when dealing with viewpoints unseen during training procedures. CNN achieves invariance via pooling operations in image classification tasks. However, the pooling operation does not necessarily improve viewpoint generalization, rather relying on more data to achieve viewpoint equivariance. Capsule network (CapsNet) is proposed to tackle this issue, but it is inefficient and inaccurate when applied to complex datasets. We propose a novel CapsNet architecture called Global Routing CapsNet (GR-CapsNet) to solve this problem. Specifically, colored background in the input image can generate invalid background voting capsules to reduce the performance of CapsNet. Therefore, we first construct a dynamic linear unit (DLU), which avoids the generation of invalid background voting capsules. Then we present two extra learnable units: frequency domain unit (FDU) and spatial unit (SPU). The former is used to capture finer features in the frequency domain and aims to improve classification performance on complex datasets. The latter is applied to construct the spatial relationship between the voting capsules and component capsules and aims to enhance robustness to affine transformation. Finally, we propose a global routing mechanism to simplify the routing process for CapsNet, which obtains more feature information to improve the performance of CapsNet. Extensive experiments on nine datasets show that our method obtains better robustness and generalization and achieves SOTA performance compared to other related methods. And it has fewer the number of parameters and GPU memory consumption than these related methods. The source code is available on https://github.com/cwpl/GR-CapsNet.

Keyword:

CapsNet

Global routing

Multi-branch

Straight-through-routing

期刊

Pattern Recognition

IF:

7.6

论文数:

1.3W

被引数:

4.5W

机构

hefei university of technology

学者数:

2.5W

论文数: 1.7W

被引数: 35

引用论文

A Pilot Study of the Efficacy of the Unified Protocol for Transdiagnostic Treatment of Emotional Disorders in Treating Posttraumatic Psychopathology: A Randomized Controlled Trial

Journal of Traumatic Stress

IF0

2021-01-16

OAAI

Meaghan L. O'Donnell; Winnie Lau; Katherine Chisholm; James Agathos; Jonathon Little; Sonia Terhaag; Rachel Brand; Andrea Putica; Alexander C. N. Holmes; Lynda Katona; Kim L. Felmingham; Kim Murray; Fardous Hosseiny; Matthew W. Gallagher

Hamstring injury prevention practices and compliance of the Nordic hamstring program in English professional football

Translational Sports Medicine

IF0

2020-11-06

OAAI

Paul Chesterton; Craig Tears; Matthew Wright; Matthew Portas

Cross-level Feature Aggregation Network for Polyp Segmentation 用于息肉分割的跨层次特征聚合网络

PATTERN RECOGNITION

IF7.6

2023-08-01

PREAI

Zhou, Tao; Zhou, Yi; He, Kelei; Gong, Chen; Yang, Jian; Fu, Huazhu; Shen, Dinggang