Using deep learning algorithm to predict the time series data with EEG and music data
碩士 === 輔仁大學 === 資訊工程學系碩士班 === 105 === This paper studies the prediction of continuous music emotion values based on physiological signals and music data. We use a two-dimensional emotion model. Valence and Arousal are predicted respectively. We use the general recurrent neural network and the more c...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Published: |
2017
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/9syhsv |
| Summary: | 碩士 === 輔仁大學 === 資訊工程學系碩士班 === 105 === This paper studies the prediction of continuous music emotion values based on physiological signals and music data. We use a two-dimensional emotion model. Valence and Arousal are predicted respectively. We use the general recurrent neural network and the more complex sequence to sequence model to do numerical prediction. Based on the EEG data, the lowest MSE of Valence is 0.0047 and the lowest MSE of Arousal is 0.0093. Based on the music data, the lowest MSE of Valence is 0.0028 and the lowest MSE of Arousal is 0.0073.
|
|---|