Multimodal Dataset Distillation via Phased Teacher Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Guo, Shengbin, Zhao, Hang, Yang, Senqiao, Jiang, Chenyang, Cheng, Yuhang, Peng, Xiangru, Shao, Rui, Tian, Zhuotao
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914424979193856
author Guo, Shengbin
Zhao, Hang
Yang, Senqiao
Jiang, Chenyang
Cheng, Yuhang
Peng, Xiangru
Shao, Rui
Tian, Zhuotao
author_facet Guo, Shengbin
Zhao, Hang
Yang, Senqiao
Jiang, Chenyang
Cheng, Yuhang
Peng, Xiangru
Shao, Rui
Tian, Zhuotao
contents Multimodal dataset distillation aims to construct compact synthetic datasets that enable efficient compression and knowledge transfer from large-scale image-text data. However, existing approaches often fail to capture the complex, dynamically evolving knowledge embedded in the later training stages of teacher models. This limitation leads to degraded student performance and compromises the quality of the distilled data. To address critical challenges such as pronounced cross-stage performance gaps and unstable teacher trajectories, we propose Phased Teacher Model with Shortcut Trajectory (PTM-ST) -- a novel phased distillation framework. PTM-ST leverages stage-aware teacher modeling and a shortcut-based trajectory construction strategy to accurately fit the teacher's learning dynamics across distinct training phases. This enhances both the stability and expressiveness of the distillation process. Through theoretical analysis and comprehensive experiments, we show that PTM-ST significantly mitigates optimization oscillations and inter-phase knowledge gaps, while also reducing storage overhead. Our method consistently surpasses state-of-the-art baselines on Flickr30k and COCO, achieving up to 13.5% absolute improvement and an average gain of 9.53% on Flickr30k. Code: https://github.com/Previsior/PTM-ST.
format Preprint
id arxiv_https___arxiv_org_abs_2603_25388
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Multimodal Dataset Distillation via Phased Teacher Models
Guo, Shengbin
Zhao, Hang
Yang, Senqiao
Jiang, Chenyang
Cheng, Yuhang
Peng, Xiangru
Shao, Rui
Tian, Zhuotao
Computer Vision and Pattern Recognition
Multimodal dataset distillation aims to construct compact synthetic datasets that enable efficient compression and knowledge transfer from large-scale image-text data. However, existing approaches often fail to capture the complex, dynamically evolving knowledge embedded in the later training stages of teacher models. This limitation leads to degraded student performance and compromises the quality of the distilled data. To address critical challenges such as pronounced cross-stage performance gaps and unstable teacher trajectories, we propose Phased Teacher Model with Shortcut Trajectory (PTM-ST) -- a novel phased distillation framework. PTM-ST leverages stage-aware teacher modeling and a shortcut-based trajectory construction strategy to accurately fit the teacher's learning dynamics across distinct training phases. This enhances both the stability and expressiveness of the distillation process. Through theoretical analysis and comprehensive experiments, we show that PTM-ST significantly mitigates optimization oscillations and inter-phase knowledge gaps, while also reducing storage overhead. Our method consistently surpasses state-of-the-art baselines on Flickr30k and COCO, achieving up to 13.5% absolute improvement and an average gain of 9.53% on Flickr30k. Code: https://github.com/Previsior/PTM-ST.
title Multimodal Dataset Distillation via Phased Teacher Models
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2603.25388