Parameters as Experts: Adapting Vision Models with Dynamic Parameter Routing

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Lou, Meng, Yu, Stanley, Yu, Yizhou
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918512496214016
author Lou, Meng
Yu, Stanley
Yu, Yizhou
author_facet Lou, Meng
Yu, Stanley
Yu, Yizhou
contents Adapting pre-trained vision models using parameter-efficient fine-tuning (PEFT) remains challenging, as it aims to achieve performance comparable to full fine-tuning using a minimal number of trainable parameters. When applied to complex dense prediction tasks, existing methods exhibit limitations, including input-agnostic modeling and redundant cross-layer representations. To this end, we propose ParaX, a new adapter-style method featuring a simple mixture-of-experts (MoE) architecture. Specifically, we introduce shared expert centers, where each expert is a trainable parameter matrix. During a feedforward pass, each ParaX module in the network dynamically generates weight matrices tailored for the current module via a simple dynamic parameter routing mechanism, which selectively aggregates parameter matrices in the corresponding expert center. Dynamic weight matrices in ParaX modules facilitate low-rank adaptation in an input-dependent manner, thus generating more customized and powerful feature representations. Moreover, since ParaX modules across multiple network layers share the same expert center, they improve feature diversity by promoting implicit cross-layer feature interaction. Extensive experimental results demonstrate the superiority of ParaX across diverse visual recognition tasks. Code is publicly released at: https://github.com/LMMMEng/ParaX.
format Preprint
id arxiv_https___arxiv_org_abs_2602_06862
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Parameters as Experts: Adapting Vision Models with Dynamic Parameter Routing
Lou, Meng
Yu, Stanley
Yu, Yizhou
Computer Vision and Pattern Recognition
Adapting pre-trained vision models using parameter-efficient fine-tuning (PEFT) remains challenging, as it aims to achieve performance comparable to full fine-tuning using a minimal number of trainable parameters. When applied to complex dense prediction tasks, existing methods exhibit limitations, including input-agnostic modeling and redundant cross-layer representations. To this end, we propose ParaX, a new adapter-style method featuring a simple mixture-of-experts (MoE) architecture. Specifically, we introduce shared expert centers, where each expert is a trainable parameter matrix. During a feedforward pass, each ParaX module in the network dynamically generates weight matrices tailored for the current module via a simple dynamic parameter routing mechanism, which selectively aggregates parameter matrices in the corresponding expert center. Dynamic weight matrices in ParaX modules facilitate low-rank adaptation in an input-dependent manner, thus generating more customized and powerful feature representations. Moreover, since ParaX modules across multiple network layers share the same expert center, they improve feature diversity by promoting implicit cross-layer feature interaction. Extensive experimental results demonstrate the superiority of ParaX across diverse visual recognition tasks. Code is publicly released at: https://github.com/LMMMEng/ParaX.
title Parameters as Experts: Adapting Vision Models with Dynamic Parameter Routing
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2602.06862