Preserving Structural Irreplaceability in Token Pruning through Local Representation Complexity
IEEE Transactions on Image Processing · 2026 · 1-1
DOI: 10.1109/tip.2026.3744689 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Image Processing, 2026. Verfasst von Chongxin Li und 2 weiteren Autor:innen. Gefördert von: National Natural Science Foundation of China, Natural Science Foundation of Shanghai.
Abstract (Originalsprache)
The effectiveness of vision token pruning critically depends on the criterion used to define token redundancy. Existing approaches, which primarily rely on global saliency scores or similarity heuristics, often fail to account for a token's essential contribution to the model's internal representations. This oversight results in retaining visually prominent but redundant tokens or merging semantically distinct features, severely degrading performance under tight compression budgets. To address this, we propose a novel token pruning strategy aligned with representation complexity. We first introduce Local Representation Complexity (LRC), a token-level measure that estimates the intrinsic dimension of each token's embedding. Tokens exhibiting high LRC are considered locally unpredictable and are therefore prioritized for preservation. To ensure selection diversity and mitigate over-concentration in the selected set, we frame the pruning process as a budget allocation problem and introduce a Greedy Selection algorithm with Penalized Diversity to actively suppress redundancy. We evaluate the proposed method across five vision-language model backbones from three model families, including LLaVA-1.5/LLaVA-NeXT, Qwen2.5-VL/Qwen3-VL, and InternVL3, covering a wide range of capacities and tasks. The results demonstrate competitive performance across different backbones, with particularly notable gains at aggressive compression ratios.
Förderung
- National Natural Science Foundation of China
- Natural Science Foundation of Shanghai
Zitieren
APA
Li, C., Lu, S., & Wang, H. (2026). Preserving Structural Irreplaceability in Token Pruning through Local Representation Complexity. IEEE Transactions on Image Processing, 1-1. https://doi.org/10.1109/tip.2026.3744689
Vancouver
Li C, Lu S, Wang H. Preserving Structural Irreplaceability in Token Pruning through Local Representation Complexity. IEEE Transactions on Image Processing. 2026:1-1. doi:10.1109/tip.2026.3744689
BibTeX
@article{Li2026,
author = {Li, Chongxin and Lu, Shiyu and Wang, Hanzhang},
title = {Preserving Structural Irreplaceability in Token Pruning through Local Representation Complexity},
journal = {IEEE Transactions on Image Processing},
year = {2026},
pages = {1--1},
doi = {10.1109/tip.2026.3744689}
}
EUVIMED