Non-Iterative Cluster Routing: Analysis and Implementation Strategies
-
Published:2024-02-20
Issue:5
Volume:14
Page:1706
-
ISSN:2076-3417
-
Container-title:Applied Sciences
-
language:en
-
Short-container-title:Applied Sciences
Author:
Pham Huong1ORCID, Cheng Samuel2ORCID
Affiliation:
1. Norman Campus, School of Electrical and Computer Engineering, University of Oklahoma, Norman, OK 73019, USA 2. Tulsa Campus, School of Electrical and Computer Engineering, University of Oklahoma, Tulsa, OK 74135, USA
Abstract
In conventional routing, a capsule network employs routing algorithms for bidirectional information flow between layers through iterative processes. In contrast, the cluster routingtechnique utilizes a non-iterative process and can outperform state-of-the-art models with fewer parameters, while preserving the part–whole relationship and demonstrating robust generalization to novel viewpoints. This paper aims to further analyze and clarify this concept, providing insights that allow users to implement the cluster routing technique efficiently. Additionally, we expand the technique and propose variations based on the routing principle of achieving consensus among votes in distinct clusters. In some cases, these variations have the potential to enhance and boost the cluster routing performance while utilizing similar memory and computing resources.
Funder
Vice President for Research and Partnerhsips the Data Institute for Societal Challenges the Stephenson Cancer Center at the University of Oklahoma
Reference41 articles.
1. Sabour, S., Frosst, N., and Hinton, G.E. (2017, January 4–9). Dynamic routing between capsules. Proceedings of the 31st Conferenceon Neural Information Processing Systems (NIPS2017), Long Beach, CA, USA. 2. Engstrom, L., Tsipras, D., Schmidt, L., and Madry, A. (2017). A rotation and a translation suffice: Fooling cnns with simple transformations. arXiv. 3. Alcorn, M.A., Li, Q., Gong, Z., Wang, C., Mai, L., Ku, W.S., and Nguyen, A. (2019, January 16–20). Strike (with) a pose: Neural networks are easily fooled by strange poses of familiar objects. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Long Beach, CA, USA. 4. Hinton, G.E., Sabour, S., and Frosst, N. (May, January 30). Matrix capsules with EM routing. Proceedings of the International Conference on Learning Representations, Vancouver, BC, Canada. 5. Hinton, G.E., Krizhevsky, A., and Wang, S.D. (2011, January 14–17). Transforming auto-encoders. Proceedings of the International Conference on Artificial Neural Networks, Espoo, Finland.
Cited by
1 articles.
订阅此论文施引文献
订阅此论文施引文献，注册后可以免费订阅5篇论文的施引文献，订阅后可以查看论文全部施引文献
|
|