A Novel Capacitive Touch Sensing Approach and Its Application in Gender Classification

Abstract

Touch sensing is widely used in electronic devices like touch screens in smartphones or touch buttons in other consumer electronics. This work aims to infer the user’s gender when they touch a pair of simple touch buttons. To realize this, we present a novel capacitive sensor design that enables complex impedance and capacitance measurement using a low-cost Vector Network Analyzer (VNA). We have developed a dedicated PC application to control VNA and manage the data collection process. Our data collection endeavors encompassed 56 male and female participants, each simultaneously interacting with the button pair. At the same time, the VNA was adjusted to measure S-parameters within the High Frequency (HF) and Very High Frequency (VHF) bands. Multiple machine learning classifiers are trained by selecting the best features from S-parameter measurement data. The Linear Support Vector Machine (SVM) classifier yielded the best performance, achieving a 5-fold cross-validation accuracy of 92.9% for gender classification. Compared to existing methods, our work demonstrates competitiveness in classification accuracy, boasting advantages in reduced complexity and cost-effectiveness.

Keywords

• Gender classification
• machine learning
• S-parameters
• touch sensing