Using ensemble classifier systems for handling missing data in emotion recognition from physiology: One step towards a practical system

Abstract

Previous work on emotion recognition from physiology has rarely addressed the problem of missing data. However, data loss due to artifacts is a frequent phenomenon in practical applications. Discarding the whole data instance if only a part is corrupted results in a substantial loss of data. To address this problem, two methods for handling missing data (imputation and reduced-feature models) in combination with two classifier fusion approaches (majority and confidence voting) are investigated in this work. The five emotions amusement, anger, contentment, neutral and sadness were elicited in 20 subjects by films while six physiological signals (ECG, EMG, EOG, EDA, respiration and finger temperature) were recorded. Results show that classifier fusion significantly increases the recognition accuracy in comparison to single classifiers by up to 16.3%. Regarding the methods for handling missing data, reduced-feature models are competitive or even slightly better than models which employ imputation. This is beneficial for practical applications where computational complexity is critical. Related workModalities which have been used to automatically detect emotions include facial expression [3-5], speech [6-8] and 978-1-4244-4799-2/09/$25.00 c 2009 IEEE

Towards long term monitoring of electrodermal activity in daily life

“…EDA is known as a relevant indicator of the emotional state and the stress level of a person [10,11]. To ensure that users accept this kind of sensing in daily life, all sensors need to be comfortable, invisible and easy to apply.…”

Signal processing technologies for activity-aware smart textiles

“…This in turn can address some of the challenges discussed above. Setz et al (2009) present a system to recognize emotions based on modalities, which could be integrated into textiles: ECG, EMG, EOG, galvanic skin response and respiration rate. Due to motion artifacts, these sensors deliver sporadicallycorrupted data.…”

Monitoring Patients in Ambulatory Palliative Care: a Design for an Observational Study

“…Emotions in addition to the wristband sensor data, emotions can be inferred from voice [22] encrypted phone calls n/a voice analysis based on features, e.g., energy of the signal, Mel Frequency Cepstral Coefficients, etc. Galvanic skin response (GSR) Impedance kOhm The GSR is used to detect stress [25].…”