Saved in:
| Main Authors: | , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2503.09445 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912302116110336 |
|---|---|
| author | Yang, Xiaoda Lu, JunYu Qiu, Hongshun Li, Sijing Li, Hao Ji, Shengpeng Tang, Xudong Xu, Jiayang Duan, Jiaqi Jiang, Ziyue Lin, Cong Cai, Sihang Xie, Zejian Song, Zhuoyang Zhang, Songxin |
| author_facet | Yang, Xiaoda Lu, JunYu Qiu, Hongshun Li, Sijing Li, Hao Ji, Shengpeng Tang, Xudong Xu, Jiayang Duan, Jiaqi Jiang, Ziyue Lin, Cong Cai, Sihang Xie, Zejian Song, Zhuoyang Zhang, Songxin |
| contents | Vision-Language Models (VLMs) based on Mixture-of-Experts (MoE) architectures have emerged as a pivotal paradigm in multimodal understanding, offering a powerful framework for integrating visual and linguistic information. However, the increasing complexity and diversity of tasks present significant challenges in coordinating load balancing across heterogeneous visual experts, where optimizing one specialist's performance often compromises others' capabilities. To address task heterogeneity and expert load imbalance, we propose Astrea, a novel multi-expert collaborative VLM architecture based on progressive pre-alignment. Astrea introduces three key innovations: 1) A heterogeneous expert coordination mechanism that integrates four specialized models (detection, segmentation, classification, captioning) into a comprehensive expert matrix covering essential visual comprehension elements; 2) A dynamic knowledge fusion strategy featuring progressive pre-alignment to harmonize experts within the VLM latent space through contrastive learning, complemented by probabilistically activated stochastic residual connections to preserve knowledge continuity; 3) An enhanced optimization framework utilizing momentum contrastive learning for long-range dependency modeling and adaptive weight allocators for real-time expert contribution calibration. Extensive evaluations across 12 benchmark tasks spanning VQA, image captioning, and cross-modal retrieval demonstrate Astrea's superiority over state-of-the-art models, achieving an average performance gain of +4.7\%. This study provides the first empirical demonstration that progressive pre-alignment strategies enable VLMs to overcome task heterogeneity limitations, establishing new methodological foundations for developing general-purpose multimodal agents. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_09445 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Astrea: A MOE-based Visual Understanding Model with Progressive Alignment Yang, Xiaoda Lu, JunYu Qiu, Hongshun Li, Sijing Li, Hao Ji, Shengpeng Tang, Xudong Xu, Jiayang Duan, Jiaqi Jiang, Ziyue Lin, Cong Cai, Sihang Xie, Zejian Song, Zhuoyang Zhang, Songxin Computer Vision and Pattern Recognition Artificial Intelligence Vision-Language Models (VLMs) based on Mixture-of-Experts (MoE) architectures have emerged as a pivotal paradigm in multimodal understanding, offering a powerful framework for integrating visual and linguistic information. However, the increasing complexity and diversity of tasks present significant challenges in coordinating load balancing across heterogeneous visual experts, where optimizing one specialist's performance often compromises others' capabilities. To address task heterogeneity and expert load imbalance, we propose Astrea, a novel multi-expert collaborative VLM architecture based on progressive pre-alignment. Astrea introduces three key innovations: 1) A heterogeneous expert coordination mechanism that integrates four specialized models (detection, segmentation, classification, captioning) into a comprehensive expert matrix covering essential visual comprehension elements; 2) A dynamic knowledge fusion strategy featuring progressive pre-alignment to harmonize experts within the VLM latent space through contrastive learning, complemented by probabilistically activated stochastic residual connections to preserve knowledge continuity; 3) An enhanced optimization framework utilizing momentum contrastive learning for long-range dependency modeling and adaptive weight allocators for real-time expert contribution calibration. Extensive evaluations across 12 benchmark tasks spanning VQA, image captioning, and cross-modal retrieval demonstrate Astrea's superiority over state-of-the-art models, achieving an average performance gain of +4.7\%. This study provides the first empirical demonstration that progressive pre-alignment strategies enable VLMs to overcome task heterogeneity limitations, establishing new methodological foundations for developing general-purpose multimodal agents. |
| title | Astrea: A MOE-based Visual Understanding Model with Progressive Alignment |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence |
| url | https://arxiv.org/abs/2503.09445 |