Issue |
ITM Web Conf.
Volume 70, 2025
2024 2nd International Conference on Data Science, Advanced Algorithm and Intelligent Computing (DAI 2024)
|
|
---|---|---|
Article Number | 02012 | |
Number of page(s) | 8 | |
Section | Machine Learning in Healthcare and Finance | |
DOI | https://doi.org/10.1051/itmconf/20257002012 | |
Published online | 23 January 2025 |
Few-shot Learning: Methods and Applications
1 Middle School Affiliated To Renmin University of China Tongzhou Campus, Beijing, China
2 Shool of Computer and Artificial Intelligence, Beijing Technology and Business University, Beijing, China
* Corresponding author: 2307010207@st.btbu.edu.cn
The Few-shot learning (FSL) approach distills meaningful features from a constrained sample set, allowing models to swiftly adjust to novel tasks and decreasing the dependency on extensive datasets. This approach leverages methods involving meta-learning, transfer learning, and data augmentation to boost the model's ability to recognize new categories. In many areas of artificial intelligence, obtaining large annotated datasets is often high in financial demand and extensively time-consuming, particularly in specialized fields with limited data availability. Therefore, the study of FSL is particularly critical. This paper first reviews the relevant methods of FSL, primarily categorizing them into model fine-tuning based FSL, data augmentation-based FSL, and transfer learning-based FSL. Model fine-tuning based FSL involves making slight adjustments to pre-trained models, allowing them to adapt to new tasks. Data augmentation-based FSL enhances the model's generalization ability by generating or expanding existing data. Transfer learning-based FSL transfers knowledge acquired by models from large datasets to smaller ones, enhancing the learning outcomes. Subsequently, this paper reviews the application areas of FSL and explores its impact in these fields. This paper aims to present the current state and prospects of this technique, providing valuable insights for researchers in related fields.
© The Authors, published by EDP Sciences, 2025
This is an Open Access article distributed under the terms of the Creative Commons Attribution License 4.0, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.