PACE: Providing Authentication through Computational Gait Evaluation with Deep learning
PACE: Providing Authentication through Computational Gait Evaluation with Deep learning
复制标题
PACE：通过深度学习计算步态评估提供身份验证
DOI：
10.1145/3565287.3617618
复制
发表时间：
2023
期刊：
影响因子：
--
通讯作者：
Youn, Jong-Hoon
中科院分区：
文献类型：
--
作者：
Rodriguez, Jesus;Youn, Jong-Hoon
这项研究提出了PACE（通过计算步态评估提供身份验证），这是一种利用深度学习算法的功能进行基于步态的身份验证的新方法。PACE的主要目标是通过利用个人表现出的独特步态模式来提高用户认证机制的安全性和效率。这项研究描述了一个深度学习模型的开发和实现，该模型是在一组提取的特征上训练的。这些特征，包括均值、方差、标准差、峰度和偏度，都来自加速度计和陀螺仪数据，作为深度学习模型的用户步态模式描述符。该模型的性能进行了评估的基础上，其分类和认证用户准确地使用这些功能的能力。为了这项研究的目的，招募了12名参与者，将传感器固定在他们的后臀部和右脚踝上，以收集必要的加速度计和陀螺仪数据。实验结果非常有希望，该模型在认证用户时达到了99%的准确率。这些发现强调了PACE作为步态认证传统机器学习方法的可行替代方案的潜力。这项研究的影响是深远的，潜在的应用程序跨越了众多的场景，其中安全性是至关重要的。
This research presents PACE (Providing Authentication through Computational Gait Evaluation), a novel methodology for gait-based authentication leveraging the power of deep learning algorithms. The primary objective of PACE is to enhance the security and efficiency of user authentication mechanisms by capitalizing on the unique gait patterns exhibited by individuals. This study delineates the development and implementation of a deep learning model, which was trained on a set of extracted features. These features, including mean, variance, standard deviation, kurtosis, and skewness, were derived from accelerometer and gyroscope data, serving as descriptors of users' gait patterns for the deep learning model. The model's performance was evaluated based on its ability to classify and authenticate users accurately using these features. For the purpose of this study, twelve participants were enlisted, with sensors affixed to their back hip and right ankle to collect the requisite accelerometer and gyroscope data. The experimental results were highly promising, with the model achieving an exceptional accuracy rate of 99% in authenticating users. These findings underscore the potential of PACE as a viable alternative to conventional machine learning methods for gait authentication. The implications of this research are far-reaching, with potential applications spanning a multitude of scenarios where security is of paramount importance.