Practical Processing of Mobile Sensor Data for Continual Deep Learning Predictions

摘要

We present a practical approach for processing mobile sensor time series datafor continual deep learning predictions. The approach comprises data cleaning,normalization, capping, time-based compression, and finally classification witha recurrent neural network. We demonstrate the effectiveness of the approach ina case study with 279 participants. On the basis of sparse sensor events, thenetwork continually predicts whether the participants would attend to anotification within 10 minutes. Compared to a random baseline, the classifierachieves a 40% performance increase (AUC of 0.702) on a withheld test set. Thisapproach allows to forgo resource-intensive, domain-specific, error-pronefeature engineering, which may drastically increase the applicability ofmachine learning to mobile phone sensor data.