Open Access
ARTICLE
Prototypical Network Based on Manhattan Distance
Zengchen Yu1, Ke Wang2,*, Shuxuan Xie1, Yuanfeng Zhong1, Zhihan Lv3
1
College of Computer Science and Technology, Qingdao University, Qingdao, China
2
Psychiatric Department, Qingdao Municipal Hospital, Qingdao, China
3
Qingdao Institute of Bioenergy and Bioprocess Technology, Chinese Academy of Sciences, Qingdao, China
* Corresponding Author: Ke Wang. Email:
Computer Modeling in Engineering & Sciences 2022, 131(2), 655-675. https://doi.org/ 10.32604/cmes.2022.019612
Received 01 October 2021; Accepted 09 November 2021; Issue published 14 March 2022
Abstract
Few-shot Learning algorithms can be effectively applied to fields where certain categories have only a small amount
of data or a small amount of labeled data, such as medical images, terrorist surveillance, and so on. The Metric
Learning in the Few-shot Learning algorithm is classified by measuring the similarity between the classified samples
and the unclassified samples. This paper improves the Prototypical Network in the Metric Learning, and changes
its core metric function to Manhattan distance. The Convolutional Neural Network of the embedded module is
changed, and mechanisms such as average pooling and Dropout are added. Through comparative experiments, it
is found that this model can converge in a small number of iterations (below 15,000 episodes), and its performance
exceeds algorithms such as MAML. Research shows that replacing Manhattan distance with Euclidean distance can
effectively improve the classification effect of the Prototypical Network, and mechanisms such as average pooling
and Dropout can also effectively improve the model.
Keywords
Cite This Article
Yu, Z., Wang, K., Xie, S., Zhong, Y., Lv, Z. (2022). Prototypical Network Based on Manhattan Distance.
CMES-Computer Modeling in Engineering & Sciences, 131(2), 655–675. https://doi.org/ 10.32604/cmes.2022.019612