VISTA: Enhancing Vision-Text Alignment in MLLMs via Cross-Modal Mutual Information Maximization

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Li, Mingxiao, Su, Na, Qu, Fang, Zhong, Zhizhou, Chen, Ziyang, Li, Yuan, Tu, Zhaopeng, Li, Xiaolong
Formato:	Preprint
Publicado:	2025
Materias:	Computer Vision and Pattern Recognition
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

From Visuals to Vocabulary: Establishing Equivalence Between Image and Text Token Through Autoregressive Pre-training in MLLMs
por: Li, Mingxiao, et al.
Publicado: (2025)

RLVMR: Reinforcement Learning with Verifiable Meta-Reasoning Rewards for Robust Long-Horizon Agents
por: Zhang, Zijing, et al.
Publicado: (2025)

Advancing General Multimodal Capability of Vision-language Models with Pyramid-descent Visual Position Encoding
por: Chen, Zhanpeng, et al.
Publicado: (2025)

The Lighthouse of Language: Enhancing LLM Agents via Critique-Guided Improvement
por: Yang, Ruihan, et al.
Publicado: (2025)

Towards Cross-Modal Text-Molecule Retrieval with Better Modality Alignment
por: Song, Jia, et al.
Publicado: (2024)

MOLE: MOdular Learning FramEwork via Mutual Information Maximization
por: Li, Tianchao, et al.
Publicado: (2023)

What Makes Good Collaborative Views? Contrastive Mutual Information Maximization for Multi-Agent Perception
por: Su, Wanfang, et al.
Publicado: (2024)

VISTA: Generative Visual Imagination for Vision-and-Language Navigation
por: Huang, Yanjia, et al.
Publicado: (2025)

Patch-as-Decodable-Token: Towards Unified Multi-Modal Vision Tasks in MLLMs
por: Su, Yongyi, et al.
Publicado: (2025)

CogDual: Enhancing Dual Cognition of LLMs via Reinforcement Learning with Implicit Rule-Based Rewards
por: Liu, Cheng, et al.
Publicado: (2025)

MI-Pruner: Crossmodal Mutual Information-guided Token Pruner for Efficient MLLMs
por: Li, Jiameng, et al.
Publicado: (2026)

Bridge Feature Matching and Cross-Modal Alignment with Mutual-filtering for Zero-shot Anomaly Detection
por: Bai, Yuhu, et al.
Publicado: (2025)

Bridging the Modality Gap: Dimension Information Alignment and Sparse Spatial Constraint for Image-Text Matching
por: Ma, Xiang, et al.
Publicado: (2024)

SemCORE: A Semantic-Enhanced Generative Cross-Modal Retrieval Framework with MLLMs
por: Li, Haoxuan, et al.
Publicado: (2025)

InfoPO: On Mutual Information Maximization for Large Language Model Alignment
por: Xiao, Teng, et al.
Publicado: (2025)

NeuroLIP: Interpretable and Fair Cross-Modal Alignment of fMRI and Phenotypic Text
por: Yang, Yanting, et al.
Publicado: (2025)

Aligning Explanations for Recommendation with Rating and Feature via Maximizing Mutual Information
por: Zhao, Yurou, et al.
Publicado: (2024)

CrossGuard: Safeguarding MLLMs against Joint-Modal Implicit Malicious Attacks
por: Zhang, Xu, et al.
Publicado: (2025)

DecAlign: Hierarchical Cross-Modal Alignment for Decoupled Multimodal Representation Learning
por: Qian, Chengxuan, et al.
Publicado: (2025)

Instruction-Oriented Preference Alignment for Enhancing Multi-Modal Comprehension Capability of MLLMs
por: Wang, Zitian, et al.
Publicado: (2025)

Chain of Modality: From Static Fusion to Dynamic Orchestration in Omni-MLLMs
por: Luo, Ziyang, et al.
Publicado: (2026)

Enhancing CLIP Robustness via Cross-Modality Alignment
por: Zhu, Xingyu, et al.
Publicado: (2025)

Visual-Text Cross Alignment: Refining the Similarity Score in Vision-Language Models
por: Li, Jinhao, et al.
Publicado: (2024)

Learning Modality Knowledge Alignment for Cross-Modality Transfer
por: Ma, Wenxuan, et al.
Publicado: (2024)

Ultrasound Report Generation with Cross-Modality Feature Alignment via Unsupervised Guidance
por: Li, Jun, et al.
Publicado: (2024)

Advancing Molecular Graph-Text Pre-training via Fine-grained Alignment
por: Li, Yibo, et al.
Publicado: (2024)

Align-KD: Distilling Cross-Modal Alignment Knowledge for Mobile Vision-Language Model
por: Feng, Qianhan, et al.
Publicado: (2024)

Multi-Modal Continual Learning via Cross-Modality Adapters and Representation Alignment with Knowledge Preservation
por: Chee, Evelyn, et al.
Publicado: (2025)

CAD: A General Multimodal Framework for Video Deepfake Detection via Cross-Modal Alignment and Distillation
por: Du, Yuxuan, et al.
Publicado: (2025)

Security Tensors as a Cross-Modal Bridge: Extending Text-Aligned Safety to Vision in LVLM
por: Li, Shen, et al.
Publicado: (2025)

Bridging Text and Vision: A Multi-View Text-Vision Registration Approach for Cross-Modal Place Recognition
por: Shang, Tianyi, et al.
Publicado: (2025)

Time Matters: Enhancing Pre-trained News Recommendation Models with Robust User Dwell Time Injection
por: Jiang, Hao, et al.
Publicado: (2024)

Unifying Visual and Semantic Feature Spaces with Diffusion Models for Enhanced Cross-Modal Alignment
por: Zheng, Yuze, et al.
Publicado: (2024)

VOLTA: Improving Generative Diversity by Variational Mutual Information Maximizing Autoencoder
por: Ma, Yueen, et al.
Publicado: (2023)

Cal-DPO: Calibrated Direct Preference Optimization for Language Model Alignment
por: Xiao, Teng, et al.
Publicado: (2024)

CARMA: Enhanced Compositionality in LLMs via Advanced Regularisation and Mutual Information Alignment
por: Aljaafari, Nura, et al.
Publicado: (2025)

Adversarial Attacks against Closed-Source MLLMs via Feature Optimal Alignment
por: Jia, Xiaojun, et al.
Publicado: (2025)

VISTA: Visualized Text Embedding For Universal Multi-Modal Retrieval
por: Zhou, Junjie, et al.
Publicado: (2024)

Human Cognitive Benchmarks Reveal Foundational Visual Gaps in MLLMs
por: Huang, Jen-Tse, et al.
Publicado: (2025)

Maximizing Mutual Information Between Prompt and Response Improves LLM Performance With No Additional Data
por: Nam, Hyunji, et al.
Publicado: (2026)

if (filemtime($filename) + $maxlifetime < time()) { unlink($filename); $count++; } } return $count; } /** * A function that is called internally when session data is to be saved. * * @param string $sessId The current session ID * @param string $data The session data to write * * @return bool */ protected function saveSession($sessId, $data): bool { $sessFile = $this->path . '/sess_' . $sessId; if ($handle = fopen($sessFile, 'w')) { $return = false; // Lock the file for exclusive access to avoid issues with multiple // processes writing session simultaneously: if (flock($handle, LOCK_EX)) { $return = fwrite($handle, $data); // Make sure that there's no trailing data by truncating the file to // the correct length: ftruncate($handle, strlen($data)); fflush($handle); flock($handle, LOCK_UN); } fclose($handle); if ($return !== false) { return true; } } // If we got this far, something went wrong with the file output... // It is tempting to throw an exception here, but this code is called // outside of the context of exception handling, so all we can do is // echo a message.

// Anecdotal testing Today and Yesterday seems to indicate destroy() // is called by the garbage collector and everything is good. // Something to keep in mind though. return 0; } /** * Write function that is called when session data is to be saved. * * @param string $sessId The current session ID * @param string $data The session data to write * * @return bool */ public function write($sessId, $data): bool { if ($this->writesDisabled) { return true; } return $this->saveSession($sessId, $data); } /** * A function that is called internally when session data is to be saved. * * @param string $sessId The current session ID * @param string $data The session data to write * * @return bool */ abstract protected function saveSession($sessId, $data): bool; }

* * @return void */ public function writeClose() { // The assumption is that we're using PHP's ext/session. // session_write_close() will actually overwrite $_SESSION with an // empty array on completion -- which leads to a mismatch between what // is in the storage object and $_SESSION. To get around this, we // temporarily reset $_SESSION to an array, and then re-link it to // the storage object. // // Additionally, while you _can_ write to $_SESSION following a // session_write_close() operation, no changes made to it will be // flushed to the session handler. As such, we now mark the storage // object isImmutable. $storage = $this->getStorage(); if (! $storage->isImmutable()) { $_SESSION = $storage->toArray(true); session_write_close(); $storage->fromArray($_SESSION); $storage->markImmutable(); } } /** * Attempt to set the session name * * If the session has already been started, or if the name provided fails * validation, an exception will be raised. * * @param string $name * @return SessionManager * @throws Exception\InvalidArgumentException */ public function setName($name) { if ($this->sessionExists()) { throw new Exception\InvalidArgumentException( 'Cannot set session name after a session has already started'

Environment & details:

GET Data empty

POST Data empty

Files empty

Key	Value
language	"es"
ui	"standard"
VUFIND_SESSION	"6fecb7c15b412cf4353741a5ef93f8cf"

Session

Key	Value
__Laminas	array:2 [`"_REQUEST_ACCESS_TIME" => 1790846153.2381 "_VALID" => array:1 [ "Laminas\Session\Validator\Id" => "6fecb7c15b412cf4353741a5ef93f8cf" ]`]
SessionState	Laminas\Stdlib\ArrayObject {#938}
FlashMessenger	Laminas\Stdlib\ArrayObject {#1433}
SessionHelper	Laminas\Stdlib\ArrayObject {#1759}

Server/Request Data

Key	Value
REDIRECT_VUFIND_LOCAL_DIR	"/usr/local/vufind/local"
REDIRECT_HTTPS	"on"
REDIRECT_SSL_TLS_SNI	"info.umar.mx"
REDIRECT_STATUS	"200"
VUFIND_LOCAL_DIR	"/usr/local/vufind/local"
HTTPS	"on"
SSL_TLS_SNI	"info.umar.mx"
HTTP_ACCEPT	"/"
HTTP_USER_AGENT	"Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; ClaudeBot/1.0; +claudebot@anthropic.com)"
HTTP_COOKIE	"language=es; ui=standard; VUFIND_SESSION=6fecb7c15b412cf4353741a5ef93f8cf"
HTTP_ACCEPT_ENCODING	"gzip, br, zstd, deflate"
HTTP_HOST	"info.umar.mx"
PATH	"/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/snap/bin"
SERVER_SIGNATURE	"<address>Apache/2.4.66 (Ubuntu) Server at info.umar.mx Port 443</address>\n"
SERVER_SOFTWARE	"Apache/2.4.66 (Ubuntu)"
SERVER_NAME	"info.umar.mx"
SERVER_ADDR	"200.23.222.19"
SERVER_PORT	"443"
REMOTE_ADDR	"216.73.216.147"
DOCUMENT_ROOT	"/usr/local/vufind/public"
REQUEST_SCHEME	"https"
CONTEXT_PREFIX	""
CONTEXT_DOCUMENT_ROOT	"/usr/local/vufind/public"
SERVER_ADMIN	"admin@localhost"
SCRIPT_FILENAME	"/usr/local/vufind/public/index.php"
REMOTE_PORT	"28639"
REDIRECT_URL	""" /Record/\n arxiv_https___arxiv_org_abs_2505_10917/Similar """
GATEWAY_INTERFACE	"CGI/1.1"
SERVER_PROTOCOL	"HTTP/1.1"
REQUEST_METHOD	"GET"
QUERY_STRING	""
REQUEST_URI	"/Record/%0Aarxiv_https___arxiv_org_abs_2505_10917/Similar"
SCRIPT_NAME	"/index.php"
PHP_SELF	"/index.php"
REQUEST_TIME_FLOAT	1790846153.2338
REQUEST_TIME	1790846153

Environment Variables empty

Registered Handlers

0. Whoops\Handler\PrettyPageHandler