无增强自我监督的人类活动识别与注意力机制和自适应时间序列混合器
Zhongwei Hou1, Jin Han2, Shixun Wu2
1The Institute of Future Civil Engineering Science and Technology, Chongqing Jiaotong University, Chongqing, China.
Annals of the New York Academy of Sciences
|March 12, 2026
概括
这项研究引入了一种用于人类活动识别 (HAR) 的新型自主监督学习 (SSL) 方法,可以避免数据增强和噪音. 新方法提高了HAR任务的准确性和可靠性,使用惯性测量单元 (IMU) 数据.
科学领域:
- 机器学习 机器学习
- 信号处理 信号处理
- 生物医学工程 生物医学工程
背景情况:
- 自主监督学习 (SSL) 通过利用未标记的数据,在人类活动识别 (HAR) 中表现出色.
- 现有的SSL方法通常严重依赖于数据增强,并可能误解噪音.
- 有限的数据集规模可能会阻碍HAR模型的准确性和实际应用.
研究的目的:
- 为HAR提出一种新的SSL目标,以减轻对数据增强的依赖,并改善噪音处理.
- 增强从惯性测量单位 (IMU) 数据中提取全球和本地特征.
- 提高HAR模型的准确性和通用性.
主要方法:
- 开发了一个新的SSL目标,集成了注意力机制和自适应时间序列混合器.
- 专注于在没有数据增强的情况下在IMU数据中捕捉全球依赖和本地特征.
- 在面具重建过程中为噪声分配较低的重量.
主要成果:
- 拟议的模型在自收集 (CQJTU-FCE) 和公共数据集 (UCI,Motion,HHAR) 上都显示出显著的改进.
- 在自收集的数据集上,实现了6.54%,8.55%和7.88%的平均准确性,F1分数和Cohen的kappa改进.
- 在公开数据集的指标中显示了平均10.63%,11.77%和13.39%的增强,证实了概括性.
结论:
- 拟议的SSL方法为HAR任务提供了更有效和可靠的解决方案.
- 该模型有效地从IMU数据中提取相关特征,同时最大限度地减少噪音的影响.
- 在各种数据集中验证了可通用性,这表明HAR在现实世界中具有强大的适用性.
相关概念视频
01:19High-Level and Low-Level Awareness
Controlled processes in human consciousness represent high-alert mental states where individuals deliberately focus their attention on achieving specific goals. Controlled processes can be seen in situations like mastering new technology, where a person might become so absorbed that they ignore surrounding distractions. Such processes involve selective attention, requiring one to concentrate on particular elements of experience while disregarding others. These are governed by executive...
01:28Automatic Processing and Automatic Social Behavior
Automatic processing refers to the cognitive operations that occur without conscious intent or awareness, playing a fundamental role in shaping social cognition and behavior. These processes enable individuals to navigate complex social environments efficiently by relying on mental shortcuts and pre-existing knowledge structures known as schemas. One of the most influential mechanisms underlying automatic processing is priming, which subtly activates mental representations through exposure to...