Cross-Modal Few-Shot Learning with Second-Order Neural Ordinary Differential Equations
We introduce SONO, a novel method leveraging Second-Order Neural Ordinary Differential Equations (Second-Order NODEs) to enhance cross-modal few-shot learning. By employing a simple yet effective architecture consisting of a Second-Order NODEs model paired with a cross-modal classifier, SONO addresses the significant challenge of overfitting, which is common in few-shot scenarios due to limited training examples. Our second-order approach can approximate a broader class of functions, enhancing the model's expressive power and feature generalization capabilities. We initialize our cross-modal classifier with text embeddings derived from class-relevant prompts, streamlining training efficiency by avoiding the need for frequent text encoder processing. Additionally, we utilize text-based image augmentation, exploiting CLIP’s robust image-text correlation to enrich training data significantly. Extensive experiments across multiple datasets demonstrate that SONO outperforms existing state-of-the-art methods in few-shot learning performance.
Authors
- Angelica I. Avilés-Rivero (ORCID: https://orcid.org/0000-0002-8878-0325)
- Carola‐Bibiane Schönlieb (ORCID: https://orcid.org/0000-0003-0099-6306)
- Zhihai He (ORCID: https://orcid.org/0000-0002-2647-8286)
- Yuyan Chen (ORCID: https://orcid.org/0009-0003-5888-5518)
- Chun-Wun Cheng (ORCID: https://orcid.org/0009-0007-3281-1202)
- Yi Zhang (ORCID: https://orcid.org/0009-0001-8026-1635)
- Junyi He
Institutions
- Harbin University of Science and Technology (CN)
- University of Electronic Science and Technology of China (CN)
- Harbin Institute of Technology (CN)
- University of Cambridge (GB)
- Fudan University (CN)
- Southern University of Science and Technology (CN)
- Tsinghua University (CN)
Publication Details
- Journal
- Proceedings of the AAAI Conference on Artificial Intelligence
- Published
- 2025-04-11
- DOI
- https://doi.org/10.1609/aaai.v39i10.33118
- Citations
- 1
- Primary Topic
- Seismic Imaging and Inversion Techniques
- Type
- article
- Field-Weighted Citation Impact
- 5.50
Funders
- National Science Foundation
- W. F. Albright Institute of Archaeological Research
- Leverhulme Trust
- Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung
- National Natural Science Foundation of China
- Tsinghua University
- Engineering and Physical Sciences Research Council
- Division of Mathematical Sciences