Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Zhong, Ziqi, Tang, Daniel
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2507.00068
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866913919860211712
author Zhong, Ziqi
Tang, Daniel
author_facet Zhong, Ziqi
Tang, Daniel
contents While multi-modal learning has advanced significantly, current approaches often treat modalities separately, creating inconsistencies in representation and reasoning. We introduce MANTA (Multi-modal Abstraction and Normalization via Textual Alignment), a theoretically-grounded framework that unifies visual and auditory inputs into a structured textual space for seamless processing with large language models. MANTA addresses four key challenges: (1) semantic alignment across modalities with information-theoretic optimization, (2) adaptive temporal synchronization for varying information densities, (3) hierarchical content representation for multi-scale understanding, and (4) context-aware retrieval of sparse information from long sequences. We formalize our approach within a rigorous mathematical framework, proving its optimality for context selection under token constraints. Extensive experiments on the challenging task of Long Video Question Answering show that MANTA improves state-of-the-art models by up to 22.6% in overall accuracy, with particularly significant gains (27.3%) on videos exceeding 30 minutes. Additionally, we demonstrate MANTA's superiority on temporal reasoning tasks (23.8% improvement) and cross-modal understanding (25.1% improvement). Our framework introduces novel density estimation techniques for redundancy minimization while preserving rare signals, establishing new foundations for unifying multimodal representations through structured text.
format Preprint
id arxiv_https___arxiv_org_abs_2507_00068
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle MANTA: Cross-Modal Semantic Alignment and Information-Theoretic Optimization for Long-form Multimodal Understanding
Zhong, Ziqi
Tang, Daniel
Computer Vision and Pattern Recognition
Artificial Intelligence
Computation and Language
While multi-modal learning has advanced significantly, current approaches often treat modalities separately, creating inconsistencies in representation and reasoning. We introduce MANTA (Multi-modal Abstraction and Normalization via Textual Alignment), a theoretically-grounded framework that unifies visual and auditory inputs into a structured textual space for seamless processing with large language models. MANTA addresses four key challenges: (1) semantic alignment across modalities with information-theoretic optimization, (2) adaptive temporal synchronization for varying information densities, (3) hierarchical content representation for multi-scale understanding, and (4) context-aware retrieval of sparse information from long sequences. We formalize our approach within a rigorous mathematical framework, proving its optimality for context selection under token constraints. Extensive experiments on the challenging task of Long Video Question Answering show that MANTA improves state-of-the-art models by up to 22.6% in overall accuracy, with particularly significant gains (27.3%) on videos exceeding 30 minutes. Additionally, we demonstrate MANTA's superiority on temporal reasoning tasks (23.8% improvement) and cross-modal understanding (25.1% improvement). Our framework introduces novel density estimation techniques for redundancy minimization while preserving rare signals, establishing new foundations for unifying multimodal representations through structured text.
title MANTA: Cross-Modal Semantic Alignment and Information-Theoretic Optimization for Long-form Multimodal Understanding
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2507.00068