Short-term emotion assessment in a recall paradigm
read more
Citations
DEAP: A Database for Emotion Analysis ;Using Physiological Signals
A Multimodal Database for Affect Recognition and Implicit Tagging
EEG-Based Emotion Recognition in Music Listening
Emotions Recognition Using EEG Signals: A Survey
Multimodal Emotion Recognition in Response to Videos
References
LIBSVM: A library for support vector machines
Pattern Recognition and Machine Learning
Pattern Recognition and Machine Learning
Related Papers (5)
Frequently Asked Questions (13)
Q2. What is the way to determine the time scales favorable to emotional assessment?
Since following the stimulus onset emotional processes in brain and peripheral signals are expected to be observable at different times, the exploration of different time resolutions is needed to determine the time scales favorable to emotional assessment from EEG and peripheral activity.
Q3. What are the drawbacks to the use of SVMs as classifiers?
There are two drawbacks to the use of SVM’s as classifiers: they are intrinsically only two-class classifiers and their output is uncalibrated so that it is not directly usable as a confidence value in the case one wants to combine outputs of different classifiers or modalities.
Q4. Why have some researchers avoided using them for others HCI applications?
Notice that because of the sensitivity of EEG sensors to noise and the fact that they often require gel to be applied on the surface of the skin, some researchers have avoided using them for others HCI applications.
Q5. How many times did the authors analyze the usability of heart rate variability?
In (Salahuddin et al., 2007) the authors analyzed the usability of heart rate variability on different time periods and concluded that 50 s of signals are necessary to accurately monitor mental stress in real settings.
Q6. What are the reasons why the elicited emotions are considered reliable?
The elicited emotions are considered reliable because (i) thinking of the same episodes ought to produce similar reactions from one trial to another, (ii) emotional episodes
Q7. What is the average accuracy for two classes?
The best average accuracy for two classes is obtained from the CP classification task with nearly 80% of well classified trials (random level at 50%), followed by the CE and NP classification tasks with respectively 78% and 74% of accuracy.
Q8. What are the main types of physiological activity measurements used to assess emotions?
Diverse types of physiological activity measurements from both the peripheral and the central nervous system have been used to assess emotions.
Q9. What is the first approach to assess human emotions?
The first approach consist in using emotion assessment as a tool for evaluating attractiveness, appreciation and user experience of software (Hazlett and Benedek, 2007).
Q10. What is the reason for rejection of trials with low confidence?
In the case the trials with low confidence are those that are misclassified such rejection should lead to an increase of accuracy.
Q11. What are the main reasons for the difficulty of recalling past episodes?
recalling past episodes and eliciting the corresponding emotions are difficult tasks and participants might need a few seconds to accomplish them.
Q12. What is the interest of fusing peripheral and EEG features for emotion assessment?
The interest of fusing peripheral and EEG features for emotion assessment was shown in (Chanel et al., 2006) through a simple concatenation of feature sets.
Q13. What is the method to compute a single covariance matrix from the complete learning set?
With theLDA it is sufficient to compute a single covariance matrix from the complete learning set without distinction between classes.