Developing a new IoT device input method that can reduce the burden on users has become an important issue. This paper proposed a system Stetho Touch that identifies touch actions using acoustic information obtained when a user's finger makes contact with a solid object. To investigate the method, we implemented a prototype of an acoustic sensing device consisting of a low-pressure melamine veneer table, a stethoscope, and an audio interface. The CNN-LSTM classification model of combining CNN and LSTM classified the five touch actions with accuracy 88.26%, f-score 87.26% in LOSO and accuracy 99.39, f-score 99.39 in 18-fold cross-validation. The contributions of this paper are the following; (1) proposed a touch action recognition method using acoustic information that is more natural and accurate than existing methods, (2) evaluated a touch action recognition method using Deep Learning that can be processed in real-time using acoustic time series raw data as input, and (3) proved the compensations for the user dependence of touch actions by providing a learning phase or performing sequential learning during use.

Objective and accurate classification of fear levels is a socially important task that contributes to developing treatments for Anxiety Disorder, Obsessive–compulsive Disorder, Post-Traumatic Stress Disorder (PTSD), and Phobia. This study examines a deep learning model to automatically estimate human fear levels with high accuracy using multichannel EEG signals and multimodal peripheral physiological signals in the DEAP dataset. The Multi-Input CNN-LSTM classification model combining Convolutional Neural Network (CNN) and Long Sort-Term Memory (LSTM) estimated four fear levels with an accuracy of 98.79% and an F1 score of 99.01% in a 10-fold cross-validation. This study contributes to the following; (1) to present the possibility of recognizing fear emotion with high accuracy using a deep learning model from physiological signals without arbitrary feature extraction or feature selection, (2) to investigate effective deep learning model structures for high-accuracy fear recognition and to propose Multi-Input CNN-LSTM, and (3) to examine the model’s tolerance to individual differences in physiological signals and the possibility of improving accuracy through additional learning.