Few-Shot Classification Based on Sparse Dictionary Meta-Learning

Zuo Jiang, Yuan Wang and Yi Tang ()
Additional contact information
Zuo Jiang: School of Mathematics and Computer Science, Yunnan Minzu University, Kunming 650500, China
Yuan Wang: School of Mathematics and Computer Science, Yunnan Minzu University, Kunming 650500, China
Yi Tang: School of Mathematics and Computer Science, Yunnan Minzu University, Kunming 650500, China

Mathematics, 2024, vol. 12, issue 19, 1-18

Abstract: In the field of Meta-Learning, traditional methods for addressing few-shot learning problems often rely on leveraging prior knowledge for rapid adaptation. However, when faced with insufficient data, meta-learning models frequently encounter challenges such as overfitting and limited feature extraction capabilities. To overcome these challenges, an innovative meta-learning approach based on Sparse Dictionary and Consistency Learning (SDCL) is proposed. The distinctive feature of SDCL is the integration of sparse representation and consistency regularization, designed to acquire both broadly applicable general knowledge and task-specific meta-knowledge. Through sparse dictionary learning, SDCL constructs compact and efficient models, enabling the accurate transfer of knowledge from the source domain to the target domain, thereby enhancing the effectiveness of knowledge transfer. Simultaneously, consistency regularization generates synthetic data similar to existing samples, expanding the training dataset and alleviating data scarcity issues. The core advantage of SDCL lies in its ability to preserve key features while ensuring stronger generalization and robustness. Experimental results demonstrate that the proposed meta-learning algorithm significantly improves model performance under limited training data conditions, particularly excelling in complex cross-domain tasks. On average, the algorithm improves accuracy by 3%.

Keywords: meta-learning; few-shot learning; sparse dictionary; consistency regularization (search for similar items in EconPapers)
JEL-codes: C (search for similar items in EconPapers)
Date: 2024
References: Add references at CitEc
Citations:

Downloads: (external link)
https://www.mdpi.com/2227-7390/12/19/2992/pdf (application/pdf)
https://www.mdpi.com/2227-7390/12/19/2992/ (text/html)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:gam:jmathe:v:12:y:2024:i:19:p:2992-:d:1485969

Access Statistics for this article

Mathematics is currently edited by Ms. Emma He

More articles in Mathematics from MDPI
Bibliographic data for series maintained by MDPI Indexing Manager ().