调查嵌入式机器学习解决方案的性能,用于对姿势行为进行分类.
Bruno Andò1, Salvatore Baglio1, Mattia Manenti1
1Department of Electrical, Electronic and Computer Engineering (DIEEI), University of Catania, 95123 Catania, Italy.
Sensors (Basel, Switzerland)
|July 30, 2025
概括
这项研究引入了一种强大的多层感知子 (MLP) 方法,用于对退行性疾病中的姿势行为进行分类. 这种方法即使在有噪音数据的情况下也表现出高可靠性,优于传统的基于值的算法.
科学领域:
- 生物医学工程 生物医学工程
- 神经学 神经学
- 机器学习 机器学习
背景情况:
- 姿势不稳定是退行性疾病的一个关键问题,经常预测跌倒.
- 监测姿势行为对于患者护理和预防跌倒至关重要.
研究的目的:
- 提出一个多层感知器 (MLP) 方法来分类四种不同的姿势行为.
- 评估MLP方法论对噪音数据和不同特征集的稳定性.
- 将MLP方法与基于值的算法进行比较.
主要方法:
- 嵌入式传感架构的实施.
- 使用多层感知器 (MLP) 进行姿势行为分类.
- 在不同噪声水平和特征集下调查数据稳定性.
主要成果:
- 在MLP方法实现了接近100%的可靠性指数,即使有杂的输入数据.
- 在所有被调查的噪音水平中,观察到的性能下降是微不足道的 (不到5%).
- 与基于值的算法相比,该方法证明了更高的稳定性.
结论:
- 开发的MLP方法提供了一个高度可靠的方法来分类姿势行为.
- 这种技术对杂的数据具有强大耐用性,使其适合于现实世界的应用.
- 这些发现表明,与现有的基于值的姿势不稳定性监测方法相比,有了显著的进步.
相关概念视频
Force Classification
2.8K
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,...
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,...
2.8K
Classification of Systems-I
742
Linearity is a system property characterized by a direct input-output relationship, combining homogeneity and additivity.
Homogeneity dictates that if an input x(t) is multiplied by a constant c, the output y(t) is multiplied by the same constant. Mathematically, this is expressed as:
Homogeneity dictates that if an input x(t) is multiplied by a constant c, the output y(t) is multiplied by the same constant. Mathematically, this is expressed as:
742


