Jove
Visualize
Contáctanos
JoVE
x logofacebook logolinkedin logoyoutube logo
ACERCA DE JoVE
Visión GeneralLiderazgoBlogCentro de Ayuda JoVE
AUTORES
Proceso de PublicaciónConsejo EditorialAlcance y PolíticasRevisión por ParesPreguntas FrecuentesEnviar
BIBLIOTECARIOS
TestimoniosSuscripcionesAccesoRecursosConsejo Asesor de BibliotecasPreguntas Frecuentes
INVESTIGACIÓN
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchivo
EDUCACIÓN
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualCentro de Recursos para ProfesoresSitio de Profesores
Términos y Condiciones de Uso
Política de Privacidad
Políticas

Videos de Conceptos Relacionados

Improving Translational Accuracy02:07

Improving Translational Accuracy

3.5K
3.5K
Improving Translational Accuracy02:07

Improving Translational Accuracy

14.0K
Base complementarity between the three base pairs of mRNA codon and the tRNA anticodon is not a failsafe mechanism. Inaccuracies can range from a single mismatch to no correct base pairing at all. The free energy difference between the correct and nearly correct base pairs can be as small as 3 kcal/ mol. With complementarity being the only proofreading step, the estimated error frequency would be one wrong amino acid in every 100 amino acids incorporated. However, error frequencies observed in...
14.0K
Modeling and Similitude01:12

Modeling and Similitude

573
Scaled modeling is a fundamental technique in engineering, enabling the study of large and complex systems by creating smaller, manageable replicas that recreate critical characteristics of the original. In hydrology and civil infrastructure, for example, scaled models of dams help analyze water flow, turbulence, and pressure. This method allows for accurate predictions of real-world behavior within a controlled environment, significantly reducing the cost and time involved in full-scale...
573
Vision01:24

Vision

59.2K
Vision is the result of light being detected and transduced into neural signals by the retina of the eye. This information is then further analyzed and interpreted by the brain. First, light enters the front of the eye and is focused by the cornea and lens onto the retina—a thin sheet of neural tissue lining the back of the eye. Because of refraction through the convex lens of the eye, images are projected onto the retina upside-down and reversed.
59.2K
Language and Cognition01:27

Language and Cognition

681
Language serves as a bridge between ideas and communication, influencing how individuals perceive and interact with the world. Psychologists have long debated whether language shapes thought or vice versa. This discussion gained grip with Edward Sapir and Benjamin Lee Whorf in the 1940s, who proposed that language determines thought, a concept known as linguistic determinism. They suggested that the vocabulary and structure of a language influence how its speakers think and perceive reality.
681
Language Development01:22

Language Development

801
Children master language quickly and with relative ease, supported by both biological predisposition and reinforcement. B. F. Skinner (1957) proposed that language is learned through reinforcement, while Noam Chomsky (1965) argued that language acquisition mechanisms are biologically determined.
The critical period for language acquisition suggests that the ability to acquire language is at its peak early in life. As people age, this proficiency decreases. Language development begins very...
801

También podría leer

Artículos Relacionados

Artículos vinculados a este trabajo por autores compartidos, revista y gráfico de citas.

Ordenar por
Same author

In vivo dedifferentiation of human epidermal cells.

Cell biology international·2007
Same author

What is in a word? No versus Yes differentially engage the lateral orbitofrontal cortex.

Emotion (Washington, D.C.)·2007
Same author

[Gene expression profile changes in oral verrucous carcinoma and oral squamous cell carcinoma].

Zhonghua kou qiang yi xue za zhi = Zhonghua kouqiang yixue zazhi = Chinese journal of stomatology·2007
Same author

Morphology of critical nuclei in solid-state phase transformations.

Physical review letters·2007
Same author

Enhanced cooperative activation effect in the hydrolytic kinetic resolution of epoxides on [Co(salen)] catalysts confined in nanocages.

Angewandte Chemie (International ed. in English)·2007
Same author

[Construction of the three-dimensional finite element model of micro -implant -maxilla].

Hua xi kou qiang yi xue za zhi = Huaxi kouqiang yixue zazhi = West China journal of stomatology·2007

Video Experimental Relacionado

Updated: Jan 8, 2026

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness
03:14

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Published on: December 6, 2024

983

Potenciación de LLMs Multimodales Fieles mediante Grounding Visual Complementario

Zheren Fu, Zhendong Mao, Lei Zhang

    IEEE transactions on image processing : a publication of the IEEE Signal Processing Society
    |December 22, 2025
    PubMed
    Resumen
    Este resumen es generado por máquina.

    Los Modelos de Lenguaje Grandes Multimodales (MLLM) pueden alucinar debido a un contexto visual insuficiente y a la deriva de la atención. Un nuevo marco de Grounding Visual Complementario (CVG) mejora la fidelidad de MLLM sin datos ni herramientas adicionales.

    Palabras clave:
    Modelos de Lenguaje Grandes MultimodalesAlucinaciónGrounding VisualConsistencia FácticaInteligencia Artificial

    Más Videos Relacionados

    Eye Tracking During Visually Situated Language Comprehension: Flexibility and Limitations in Uncovering Visual Context Effects
    07:36

    Eye Tracking During Visually Situated Language Comprehension: Flexibility and Limitations in Uncovering Visual Context Effects

    Published on: November 30, 2018

    16.3K

    Videos de Experimentos Relacionados

    Last Updated: Jan 8, 2026

    Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness
    03:14

    Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

    Published on: December 6, 2024

    983
    Eye Tracking During Visually Situated Language Comprehension: Flexibility and Limitations in Uncovering Visual Context Effects
    07:36

    Eye Tracking During Visually Situated Language Comprehension: Flexibility and Limitations in Uncovering Visual Context Effects

    Published on: November 30, 2018

    16.3K

    Área de la Ciencia:

    • Inteligencia Artificial
    • Visión por Computadora
    • Procesamiento del Lenguaje Natural

    Sus antecedentes:

    • Los Modelos de Lenguaje Grandes Multimodales (MLLM) se destacan en tareas de visión y lenguaje, pero sufren de alucinaciones, generando texto inconsistente con la entrada visual.
    • Los métodos actuales para mitigar las alucinaciones son a menudo ad hoc, intensivos en datos o computacionalmente costosos.
    • Los factores clave que contribuyen a la alucinación de MLLM incluyen el contexto visual insuficiente y la deriva textual progresiva durante la generación.

    Objetivo del estudio:

    • Abordar los problemas centrales de contexto visual insuficiente y deriva textual progresiva en MLLMs.
    • Introducir un marco novedoso, Complementary Visual Grounding (CVG), que mejora la consistencia fáctica en las salidas de MLLM.
    • Desarrollar un método que aproveche la arquitectura intrínseca de los MLLM sin dependencias externas.

    Principales métodos:

    • El marco propuesto de Grounding Visual Complementario (CVG) descompone el contexto visual en ramas complementarias relevantes para la consulta.
    • CVG asegura un grounding visual sostenido durante el proceso de generación de texto autorregresivo.
    • Contrasta las distribuciones de salida de estas ramas para generar respuestas factualmente fundamentadas.

    Principales resultados:

    • CVG logra un rendimiento de vanguardia en puntos de referencia de alucinación.
    • El marco demuestra efectividad en varias arquitecturas y escalas de MLLM.
    • Los experimentos confirman la capacidad de CVG para producir respuestas fieles manteniendo el grounding visual.

    Conclusiones:

    • El Grounding Visual Complementario (CVG) mitiga eficazmente las alucinaciones en los Modelos de Lenguaje Grandes Multimodales (MLLM).
    • El marco CVG ofrece una solución intrínseca y eficiente al mejorar la utilización del contexto visual y prevenir la deriva textual.
    • Este enfoque avanza la confiabilidad y la precisión fáctica de los MLLM en tareas de visión y lenguaje.