InfoTok: Adaptive Discrete Video Tokenizer via Information-Theoretic Compression

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ye, Haotian, He, Qiyuan, Han, Jiaqi, Li, Puheng, Fan, Jiaojiao, Hao, Zekun, Reda, Fitsum, Balaji, Yogesh, Chen, Huayu, Liu, Sheng, Yao, Angela, Zou, James, Ermon, Stefano, Wang, Haoxiang, Liu, Ming-Yu
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917356000772096
author Ye, Haotian
He, Qiyuan
Han, Jiaqi
Li, Puheng
Fan, Jiaojiao
Hao, Zekun
Reda, Fitsum
Balaji, Yogesh
Chen, Huayu
Liu, Sheng
Yao, Angela
Zou, James
Ermon, Stefano
Wang, Haoxiang
Liu, Ming-Yu
author_facet Ye, Haotian
He, Qiyuan
Han, Jiaqi
Li, Puheng
Fan, Jiaojiao
Hao, Zekun
Reda, Fitsum
Balaji, Yogesh
Chen, Huayu
Liu, Sheng
Yao, Angela
Zou, James
Ermon, Stefano
Wang, Haoxiang
Liu, Ming-Yu
contents Accurate and efficient discrete video tokenization is essential for long video sequences processing. Yet, the inherent complexity and variable information density of videos present a significant bottleneck for current tokenizers, which rigidly compress all content at a fixed rate, leading to redundancy or information loss. Drawing inspiration from Shannon's information theory, this paper introduces InfoTok, a principled framework for adaptive video tokenization. We rigorously prove that existing data-agnostic training methods are suboptimal in representation length, and present a novel evidence lower bound (ELBO)-based algorithm that approaches theoretical optimality. Leveraging this framework, we develop a transformer-based adaptive compressor that enables adaptive tokenization. Empirical results demonstrate state-of-the-art compression performance, saving 20% tokens without influence on performance, and achieving 2.3x compression rates while still outperforming prior heuristic adaptive approaches. By allocating tokens according to informational richness, InfoTok enables a more compressed yet accurate tokenization for video representation, offering valuable insights for future research.
format Preprint
id arxiv_https___arxiv_org_abs_2512_16975
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle InfoTok: Adaptive Discrete Video Tokenizer via Information-Theoretic Compression
Ye, Haotian
He, Qiyuan
Han, Jiaqi
Li, Puheng
Fan, Jiaojiao
Hao, Zekun
Reda, Fitsum
Balaji, Yogesh
Chen, Huayu
Liu, Sheng
Yao, Angela
Zou, James
Ermon, Stefano
Wang, Haoxiang
Liu, Ming-Yu
Computer Vision and Pattern Recognition
Artificial Intelligence
Accurate and efficient discrete video tokenization is essential for long video sequences processing. Yet, the inherent complexity and variable information density of videos present a significant bottleneck for current tokenizers, which rigidly compress all content at a fixed rate, leading to redundancy or information loss. Drawing inspiration from Shannon's information theory, this paper introduces InfoTok, a principled framework for adaptive video tokenization. We rigorously prove that existing data-agnostic training methods are suboptimal in representation length, and present a novel evidence lower bound (ELBO)-based algorithm that approaches theoretical optimality. Leveraging this framework, we develop a transformer-based adaptive compressor that enables adaptive tokenization. Empirical results demonstrate state-of-the-art compression performance, saving 20% tokens without influence on performance, and achieving 2.3x compression rates while still outperforming prior heuristic adaptive approaches. By allocating tokens according to informational richness, InfoTok enables a more compressed yet accurate tokenization for video representation, offering valuable insights for future research.
title InfoTok: Adaptive Discrete Video Tokenizer via Information-Theoretic Compression
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2512.16975