Frequency-Aware Token Reduction for Efficient Vision Transformer

Dong-Jae Lee, Jiwan Hur, Jaehyun Choi, Jaemyung Yu, Junmo Kim

Keywords: Efficient Transformer, Token pruning, Token merging

TL;DR: Token reduction with mitigating rank-collapsing (over-smoothing) through high-frequency token preservation

Abstract:

Vision Transformers have demonstrated exceptional performance across various computer vision tasks, yet their quadratic computational complexity concerning token length remains a significant challenge. To address this, token reduction methods have been widely explored. However, existing approaches often overlook the frequency characteristics of self-attention, such as rank collapsing and over-smoothing phenomenon.
In this paper, we propose a frequency-aware token reduction strategy that improves computational efficiency while preserving performance by mitigating rank collapsing. Our method partitions tokens into high-frequency tokens and low-frequency tokens. high-frequency tokens are selectively preserved, while low-frequency tokens are aggregated into a compact direct current token to retain essential low-frequency components.
Through extensive experiments and analysis, we demonstrate that our approach significantly improves accuracy while reducing computational overhead and mitigating rank collapsing and over smoothing. Furthermore, we analyze the previous methods, shedding light on their implicit frequency characteristics and limitations. The code is available in https://github.com/jhtwosun/frequency-aware-token-pruning.

Primary Area: Deep learning (e.g., architectures, generative models, optimization for deep networks, foundation models, LLMs)

Submission Number: 2010