EMCompress: Video-LLMs with Endomorphic Multimodal Compression

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Fan, Zheyu, Liu, Jiateng, Zhang, Yuji, Wang, Zihan, Fung, Yi R., Li, Manling, Ji, Heng
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866913060663328768
author Fan, Zheyu
Liu, Jiateng
Zhang, Yuji
Wang, Zihan
Fung, Yi R.
Li, Manling
Ji, Heng
author_facet Fan, Zheyu
Liu, Jiateng
Zhang, Yuji
Wang, Zihan
Fung, Yi R.
Li, Manling
Ji, Heng
contents Video-LLMs face a fundamental tension in long-video reasoning: static, sparse frame sampling either dilutes evidence across task-irrelevant segments at significant cost or misses fine-grained temporal semantics altogether. We propose a novel, cognitively-inspired task -- Endomorphic Multimodal Compression (EMC) -- as a structurally-constrained sufficient-statistic problem for VideoQA, and formulate it as an endomorphic transformation F_EMC : (V, Q) -> (v, q) that compresses the multimodal input while preserving answer invariance across reasonable downstream models. The endomorphic form keeps the compressed output in the downstream pipeline's native task space -- a structural mirror of the filter-then-reason mechanism in the cognitive literature motivating EMC -- distinguishing it from latent-code compression (IB / VIB) and making the formulation extensible to other multimodal settings. Under the Markov chain A -> (V, Q) -> (v, q), EMC realizes the classical sufficiency condition I((v, q); A) = I((V, Q); A) in its VideoQA-natural form. As a modular front-end, EMC plugs into both Video Instruction Tuning and Video Question Answering pipelines. We release the first dedicated benchmark and propose ReSimplifyIt, an EMC baseline surpassing prior methods by 0.40 F-1 with competitive query rewriting. Integrating EMC yields relative gains of 7.33% in training and 33.7% in inference for video-language understanding.
format Preprint
id arxiv_https___arxiv_org_abs_2508_21094
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle EMCompress: Video-LLMs with Endomorphic Multimodal Compression
Fan, Zheyu
Liu, Jiateng
Zhang, Yuji
Wang, Zihan
Fung, Yi R.
Li, Manling
Ji, Heng
Computer Vision and Pattern Recognition
Video-LLMs face a fundamental tension in long-video reasoning: static, sparse frame sampling either dilutes evidence across task-irrelevant segments at significant cost or misses fine-grained temporal semantics altogether. We propose a novel, cognitively-inspired task -- Endomorphic Multimodal Compression (EMC) -- as a structurally-constrained sufficient-statistic problem for VideoQA, and formulate it as an endomorphic transformation F_EMC : (V, Q) -> (v, q) that compresses the multimodal input while preserving answer invariance across reasonable downstream models. The endomorphic form keeps the compressed output in the downstream pipeline's native task space -- a structural mirror of the filter-then-reason mechanism in the cognitive literature motivating EMC -- distinguishing it from latent-code compression (IB / VIB) and making the formulation extensible to other multimodal settings. Under the Markov chain A -> (V, Q) -> (v, q), EMC realizes the classical sufficiency condition I((v, q); A) = I((V, Q); A) in its VideoQA-natural form. As a modular front-end, EMC plugs into both Video Instruction Tuning and Video Question Answering pipelines. We release the first dedicated benchmark and propose ReSimplifyIt, an EMC baseline surpassing prior methods by 0.40 F-1 with competitive query rewriting. Integrating EMC yields relative gains of 7.33% in training and 33.7% in inference for video-language understanding.
title EMCompress: Video-LLMs with Endomorphic Multimodal Compression
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2508.21094