心拍変動解析と嗜好におけるCNNを用いた多数決法の検討

抄録

In order to provide products and services that are optimized to the values of each consumer, biometric information such as heart rate variability and electroencephalogram is useful for estimating values and preferences from the consumer’s unconscious. In this research, we create a heart rate variability spectrogram (HRVS) of a subject for a still image, and create a scalogram by using continuous wavelet transform in addition to the temporal differential change of the spectrogram. In addition, a scalogram is created using the continuous wavelet transform. From these, we consider whether human preferences can be discriminated using convolutional neural network (CNN) and support vector machine (SVM).