This study aims to determine the optimal electrode positions and re-referencing strategies for ear biosignals by systematically comparing ear and conventional biosignals. We analysed four physiological signals: electroencephalography (EEG), electromyography (EMG), electrooculography (EOG), and electrocardiography (ECG), with five re-referencing strategies. The signals were recorded from conventional locations and around the ears. Quantitative measures, such as the signal-to-noise ratio, F1 score, and correlation, were employed to identify the optimal electrode positions and re-referencing strategies for each physiological signal type. The optimal ear electrode positions were selected based on their proximity to conventional configurations: the upper part/behind the ear for EEG, the lower part for blink detection and vertical EOG, and the front for clench EMG and horizontal EOG. While no single re-referencing strategy consistently performed best for all physiological signals, we found that (mean-)ipsilateral re-referencing strategies were more suitable for EEG, EMG, blink detection, and vertical EOG, whereas (mean-)contralateral re-referencing strategies were more effective for horizontal EOG and ECG compared to other strategies. The optimal ear electrode positions and re-referencing methods varied slightly across subjects and demonstrated comparable performance to the conventional configurations in terms of quantitative measures. Ear biosignals have the potential for use in developing human-computer interfaces with comparable performance to conventional electrode configuration while being more practical in terms of attachment and detachment. This study provides in-depth insight into the optimal electrode locations and re-referencing strategy for developing ear-biosignal-based human-computer-interface applications.

This study investigates enhancing the cosmetic selection process in VR/AR environments by focusing on tactile feedback. While VR/AR allows users to visually select cosmetic shades, the lack of tactile feedback is a limitation. We used EEG and machine learning to analyze emotional and tactile responses to different cosmetic formulations and classify user preferences. The beta frequency band achieved the highest classification accuracy at 71.84%, demonstrating EEG’s potential to capture subtle emotional responses. These findings suggest that integrating tactile feedback into VR/AR could improve the realism and satisfaction of the cosmetic selection experience.

To enhance the sense of reality in the VR environment, it is essential to incorporate not only visual and hearing but also tactile feedback. Previous studies related to tactile sensation have only focused on object classification to identify the object in contact with the tactile sensor. When people encounter a new object, they can use tactile perception information, such as roughness, to understand the object’s attributes and functions. Therefore, for tactile research to be applicable in real-life scenarios, tactile perception classification is essential not only for classifying an object but also for recognizing its attributes and functions. This study aims to investigate the possibility of predicting roughness, one of the important human tactile perceptions, using deep learning models. We created 63 tactile samples, measured the vibration data of each sample using an acceleration sensor, and investigated how humans perceive the roughness of the tactile samples with five different levels. Object classification was performed using five deep learning models, among which the Shallow ConvNet achieved the highest performance, confirming its suitability as the most appropriate model for analyzing vibration data. Using the Shallow ConvNet, we attained a high accuracy of $\mathbf{9 8. 6} \pm \mathbf{5. 6 \%}$ for the identification of five levels of roughness, demonstrating that different levels of tactile perception information can be discriminated with high performance.

Biometrics based on electromyogram (EMG) and inertial measurement unit (IMU) signals has received significant attention due to its resistance to forgery and tampering, as well as its relatively higher convenience and security compared to other biometric modalities. However, most EMG and IMU signal-based authentication methods require explicit authentication using unnatural gestures, which is not userfriendly. To address this, we collected EMG and IMU data during virtual reality (VR) rhythm game play, which does not require explicit gestures, and performed implicit authentication. Using a linear discriminant analysis (LDA) classifier, we achieved classification accuracies of 91.71% for human identification and 87.91% for human authentication. The result demonstrated the feasibility of a seamless and implicit authentication in VR environments.

Brain–computer interface (BCI) technology has been studied with the fundamental goal of helping disabled people communicate with the outside world using brain signals. In particular, a large body of research has been reported in the electroencephalography (EEG)-based BCI research field during recent years. To provide a thorough summary of recent research trends in EEG-based BCIs, the present study reviewed BCI research articles published from 2007 to 2011 and investigated (a) the number of published BCI articles, (b) BCI paradigms, (c) aims of the articles, (d) target applications, (e) feature types, (f) classification algorithms, (g) BCI system types, and (h) nationalities of the author. The detailed survey results are presented and discussed one by one. [Supplemental materials are available for this article. Go to the publisher's online edition of International Journal of Human-Computer Interaction to view the free supplemental file: Supplementary Tables.pdf.]