.png){kind=icon}
.png){kind=icon}
Return
Efficient token pruning in Vision Transformers using an attention-based Multilayer Network
.png){kind=icon}
OA
AI
DOI：10.1016/j.eswa.2025.127449.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Vision Transformers (ViTs), although very successful, have a major limitation to overcome, namely the need significant computational resources to use them. Several approaches have been proposed to limit the resources required to work with ViTs, aiming at pruning the data provided in input to them. In this paper, we propose Token Reduction via an Attention-based Multilayer network (TRAM), the first approach that achieves goal using a multilayer network-based representation of the attention matrices. TRAM can work with ViTs without the need for fine-tuning. It makes several contributions to the literature in this research area; particular, it is characterized by: (i) a new representation of ViTs based on a multilayer network; (ii) a approach to evaluate the relevance of tokens based on a new centrality measure computed on the multilayer network; and (iii) an approach to reduce the number of tokens based on this centrality measure. We validated TRAM by comparing it with several state-of-the-art approaches during an extensive experimental campaign carried out on different image datasets. The results obtained demonstrate not only the efficiency but also the effectiveness of TRAM in reducing the computational load of ViTs while still allowing them provide accurate results.
Keywords:
Token relevance centrality
Multilayer network
Token pruning
Vision transformer
Computer vision
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
7.5
Papers:
2.9W
Citations:
10.2W
Organization
No organization information available