Search research articles

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 Journal Methods Collections JoVE Encyclopedia of Experiments 存档

教育

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual 教师资源中心教师网站

使用条款与条件

相关概念视频

Force Classification

Force Classification

Forces play a crucial role in the study of physics and engineering. They are essential in describing the motion, behavior, and equilibrium of objects in the physical world. Forces can be classified based on their origin, type, and direction of action.
Contact and non-contact forces are two of the most widely used categories of forces. As the name suggests, contact forces require physical contact between two objects to act upon each other. Examples of contact forces include frictional,...

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

Beyond the Bedside: Machine Learning-Guided Length of Stay (LOS) Prediction for Cardiac Patients in Tertiary Care.

Healthcare (Basel, Switzerland)·2024

查看所有相关文章

Search research articles

相关实验视频

Updated: Jun 14, 2025

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

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

Published on: December 15, 2023

用更少的资源实现更多:用于高级人类活动识别 (HAR) 的轻量级深度学习解决方案.

Sarab AlMuhaideb¹, Lama AlAbdulkarim¹, Deemah Mohammed AlShahrani¹

¹Department of Computer Science, College of Computer and Information Sciences, King Saud University, P.O. Box 266, Riyadh 11362, Saudi Arabia.

Sensors (Basel, Switzerland)

|August 29, 2024

概括

这项研究用剩余LSTM和数据增强取代了人类活动识别模型中的计算密集型BiLSTM层. 剩余LSTM方法提供了可比的精度,显著降低了模型复杂性,以实现高效的部署.

关键词:

惯性测量单位是一种惯性测量单位.卷积神经网络 (CNN) 是一种神经网络.数据增强数据增强深度学习是一种深度学习.人类活动识别 (HAR)轻量级的模型轻量级的模型.长时间的短期记忆 (LSTM)传感器数据传感器数据

更多相关视频

A Step-by-Step Implementation of DeepBehavior, Deep Learning Toolbox for Automated Behavior Analysis

A Step-by-Step Implementation of DeepBehavior, Deep Learning Toolbox for Automated Behavior Analysis

Published on: February 6, 2020

Deep Neural Networks for Image-Based Dietary Assessment

Deep Neural Networks for Image-Based Dietary Assessment

Published on: March 13, 2021

相关实验视频

Last Updated: Jun 14, 2025

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

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

Published on: December 15, 2023

A Step-by-Step Implementation of DeepBehavior, Deep Learning Toolbox for Automated Behavior Analysis

A Step-by-Step Implementation of DeepBehavior, Deep Learning Toolbox for Automated Behavior Analysis

Published on: February 6, 2020

Deep Neural Networks for Image-Based Dietary Assessment

Deep Neural Networks for Image-Based Dietary Assessment

Published on: March 13, 2021

科学领域:

计算机科学计算机科学
人工智能的人工智能
机器学习机器学习

背景情况:

人类活动识别 (HAR) 对医疗保健,健身和国防的应用至关重要.
深度学习模型,特别是双向长期短期存储器 (BiLSTM) 网络,在HAR中表现出色,但对资源有限的设备来说计算要求很高.
BiLSTMs的高参数数和计算复杂性限制了它们在智能手机等设备上的实时应用.

研究的目的:

在HAR模型中研究近似计算密集型BiLSTM组件.
探索替代模型组件和数据增强技术,以减少计算负载.
在资源有限的平台上保持或改进HAR性能.

主要方法:

用剩余LSTM (ResLSTM) 和标准LSTM取代BiLSTM层,其中包括卷积网络.
使用数据翻转增强来补偿因删除BiLSTM而失去的上下文意识.
在基准数据集上评估了修改后的架构:UCI-HAR,WISDM和KU-HAR.

主要成果:

在数据集中,ResLSTM模型表现出比BiLSTM更高或可比的准确性.
与BiLSTM相比,ResLSTM实现了显著较低的参数数量和计算复杂性.
在UCI-HAR上,ResLSTM以576,702个参数达到96.34%的准确性,而BiLSTM的95.22%为849,534.
在WISDM上,ResLSTM以192,238个参数实现了97.20%的精度,而BiLSTM的97.23%为283,182.2.
在KU-HAR上,ResLSTM获得了97.05%的准确性,386,038个参数与BiLSTM的98.63%的准确性相比,569,462.

结论:

剩余的LSTM网络为HAR提供了BiLSTM的高效替代方案,保持高精度,降低计算需求.
拟议的方法可以在智能手机和其他边缘设备上实际部署先进的HAR模型.
数据增强技术可以有效地补充简单的架构来复制复杂的模型功能.