Jove
Visualize
联系我们
JoVE
x logofacebook logolinkedin logoyoutube logo
关于 JoVE
概览领导团队博客JoVE 帮助中心
作者
出版流程编辑委员会范围与政策同行评审常见问题投稿
图书馆员
用户评价订阅访问资源图书馆顾问委员会常见问题
研究
JoVE JournalMethods CollectionsJoVE Encyclopedia of Experiments存档
教育
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab Manual教师资源中心教师网站
使用条款与条件
隐私政策
政策

相关概念视频

Causes of Similarity-Dissimilarity Effect01:26

Causes of Similarity-Dissimilarity Effect

254
The similarity-dissimilarity effect, a fundamental concept in social psychology, explains how interpersonal similarities and differences influence attraction and social interactions. This effect is supported by three key psychological perspectives: balance theory, social comparison theory, and consensual validation.Balance Theory and Cognitive ConsistencyBalance theory, developed by Fritz Heider, posits that individuals seek cognitive consistency in their relationships. When two people share...
254

您也可能阅读

相关文章

通过共同作者、期刊和引用图与本文相关的文章。

排序
Same author

A Transformer-Based Variational Autoencoder for Training Data Generation in Spindle Motor Vibration-Based Anomaly Detection.

Sensors (Basel, Switzerland)·2026
Same author

Multi-angle beam range measurement framework for carbon-ion radiotherapy using a commercial multi-layer ionization chamber.

Zeitschrift fur medizinische Physik·2026
Same author

Precision Biomarker Identification in Gynecological Cancers Using Coexpression Networks and Attention-Based LSTM in Healthcare 4.0.

Diagnostics (Basel, Switzerland)·2026
Same author

Immunological response of using emulsifying elongation factor 1 alpha of Plasmodium falciparum-based vaccines in complete Freund's adjuvant.

Vaccine·2026
Same author

Respiratory Muscle Training Combinations in Amateur Runners: A Randomized Trial of Pulmonary Function, Respiratory Muscle Strength, and Exercise Capacity.

Bioengineering (Basel, Switzerland)·2026
Same author

Dipole-Driven Charge Trapping in Monolayer Janus MoSSe for Ultrathin Nonvolatile Memory Devices.

Nano-micro letters·2026
Same journal

Intervention Feasible Region and Driver Risk Capacity Aware Human-Machine Collaborative Safe Trajectory Planning.

IEEE transactions on neural networks and learning systems·2026
Same journal

A Unified Differential Denoising Learning Framework With a Pre-Trained Model and Fuzzy Graph Networks for Drug-Drug Interaction Prediction.

IEEE transactions on neural networks and learning systems·2026
Same journal

Self-Supervised Continuous Dynamic Graph Representation Learning via Hawkes Processes.

IEEE transactions on neural networks and learning systems·2026
Same journal

cPU: Consistent Risk Estimator for Positive-Unlabeled Learning.

IEEE transactions on neural networks and learning systems·2026
Same journal

Tuning-Free Latent Diffusion Models for Ultrahigh-Resolution Image Editing.

IEEE transactions on neural networks and learning systems·2026
Same journal

Hidden Data Recovery and Forecasting via Next-Generation Reservoir Computing With Multiscale Delay Selection.

IEEE transactions on neural networks and learning systems·2026
查看所有相关文章

相关实验视频

Updated: Jan 17, 2026

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
06:37

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention

Published on: December 15, 2023

5.3K

亚马普:通过基于相似性的修剪指标自动多头注意力修剪.

Eunho Lee, Youngbae Hwang

    IEEE transactions on neural networks and learning systems
    |September 15, 2025
    PubMed
    概括
    此摘要是机器生成的。

    我们在视觉变压器中引入了用于线性注意力的自动修剪方法,显著降低了计算成本,同时保持了性能. 这种方法解决了多头注意力中的道不匹配问题,从而实现了高效的模型优化.

    更多相关视频

    Measuring Attention and Visual Processing Speed by Model-based Analysis of Temporal-order Judgments
    13:00

    Measuring Attention and Visual Processing Speed by Model-based Analysis of Temporal-order Judgments

    Published on: January 23, 2017

    10.3K
    Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique
    04:48

    Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

    Published on: July 5, 2024

    736

    相关实验视频

    Last Updated: Jan 17, 2026

    Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
    06:37

    Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention

    Published on: December 15, 2023

    5.3K
    Measuring Attention and Visual Processing Speed by Model-based Analysis of Temporal-order Judgments
    13:00

    Measuring Attention and Visual Processing Speed by Model-based Analysis of Temporal-order Judgments

    Published on: January 23, 2017

    10.3K
    Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique
    04:48

    Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

    Published on: July 5, 2024

    736

    科学领域:

    • 计算机视觉 计算机视觉
    • 机器学习 机器学习
    • 人工智能的人工智能

    背景情况:

    • 变压器在视觉任务中实现了高性能,但在自我注意中遭受二次计算复杂性的困扰.
    • 线性注意力提供了线性复杂性的替代方案,改善了计算性能的权衡.
    • 自动修剪优化模型结构的资源限制,但面临的挑战是多头注意力通道不匹配.

    研究的目的:

    • 为视觉变压器中的线性注意力机制提出一种自动修剪方法,以解决频道不匹配问题.
    • 为了提高视觉任务的变压器模型的计算效率和性能.

    主要方法:

    • 整合了基于频道相似性的权重,用于修剪指标,以保持每个注意力头部内的信息频道.
    • 调整了修剪指示器,以确保在所有头部均地移除通道,防止通道不匹配.
    • 整合了重权模块以减轻信息丢失,并引入了有效的修剪指示器初始化,以获得线性注意力.

    主要成果:

    • 在ImageNet-1K上的FLattenTransformer的FLOPs (每秒浮点操作) 减少了30%,性能几乎没有损失.
    • 与DeiT-B模型相比,表现出1.96%的精度增长,同时减少了37%的FLOP.
    • 与Swin-B模型相比,显示了1.05%的精度增加,FLOP减少了10%.

    结论:

    • 提出的自动修剪方法有效地优化视觉变压器中的线性注意力机制.
    • 该方法成功地解决了通道不匹配问题,从而在不影响性能的情况下节省了大量的计算成本.
    • 在效率和准确性方面,超越现有的最先进的高效模型和最新的修剪技术.