---
title: "Connecting Giants: Synergistic Knowledge Transfer of Large Multimodal Models for Few-Shot Learning"
authors:
  - "Hao Tang"
  - "Shengfeng He"
  - "Jing Qin"
doi: "10.24963/ijcai.2025/693"
url: "https://ijcai.org/proceedings/2025/693"
pdf: "https://ijcai.org/proceedings/2025/0693.pdf"
venue: "Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence"
year: 2025
pages: "6227-6235"
track: "Main Track"
keywords:
  - "Machine Learning: ML: Meta-learning"
  - "Machine Learning: ML: Few-shot learning"
  - "Computer Vision: CV: Transfer, low-shot, semi- and un- supervised learning"
  - "Computer Vision: CV: Multimodal learning"
---

# Connecting Giants: Synergistic Knowledge Transfer of Large Multimodal Models for Few-Shot Learning

Hao Tang, Shengfeng He, Jing Qin

*Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence*, pages 6227-6235. DOI: [10.24963/ijcai.2025/693](https://doi.org/10.24963/ijcai.2025/693)

## Abstract

Few-shot learning (FSL) addresses the challenge of classifying novel classes with limited training samples. While some methods leverage semantic knowledge from smaller-scale models to mitigate data scarcity, these approaches often introduce noise and bias due to the data’s inherent simplicity. In this paper, we propose a novel framework, Synergistic Knowledge Transfer (SynTrans), which effectively transfers diverse and complementary knowledge from large multimodal models to empower the off-the-shelf few-shot learner. Specifically, SynTrans employs CLIP as a robust teacher and uses a few-shot vision encoder as a weak student, distilling semantic-aligned visual knowledge via an unsupervised proxy task. Subsequently, a training-free synergistic knowledge mining module facilitates collaboration among large multimodal models to extract high-quality semantic knowledge. Building upon this, a visual-semantic bridging module enables bi-directional knowledge transfer between visual and semantic spaces, transforming explicit visual and implicit semantic knowledge into category-specific classifier weights. Finally, SynTrans introduces a visual weight generator and a semantic weight reconstructor to adaptively construct optimal multimodal FSL classifiers. Experimental results on four FSL datasets demonstrate that SynTrans, even when paired with a simple few-shot vision encoder, significantly outperforms current state-of-the-art methods.