基于机器学习的多用户触觉手势识别研究

Research on Multi-user Tactile Gesture Recognition Based on Machine Learning

触觉手势数据分类对身份认证具有非常重要的意义。本文简要介绍了k NN、SVM、ELM三种分类方法的基本原理,采用公开可用的Touchalytics触觉手势数据集,利用三种分类方法对多类触觉手势数据进行分类。实验结果表明极限学习机(ELM)相较其他分类方法,测试时间合理,识别准确率高,能够有效实现触觉手势分类识别。

Tactile gesture data classification is very important for identity authentication. In this paper, we briefly introduce the basic principles of kNN,SVM and ELM, and then use the three classification methods for tactile gesture recognition. Experiments using publicly available screen touch dataset -Touchalytics dataset show that extreme learning machine （ELM） compares favorably to other classification methods. The recognition accuracy is good and the testing time is reasonable. It can effectively achieve tactile gesture classification and recognition.