Academia.eduAcademia.edu

Figure 3 – uploaded by Yi-hsuan Yang

See full PDFdownloadDownload figure
Table 3 | The informative EEG features that consistently appeared across multiple subjects. competed to the EEG features and replaced the ones with rela- tively low discriminative power, especially for arousal scale. This evidently explains the reason that the subject-independent mul- timodal approach leading to significant improvements upon the subject-independent EEG results. Table 4 lists these informative musical features, which consistently appeared in above half of the subjects.

Table 3 | The informative EEG features that consistently appeared across multiple subjects. competed to the EEG features and replaced the ones with rela- tively low discriminative power, especially for arousal scale. This evidently explains the reason that the subject-independent mul- timodal approach leading to significant improvements upon the subject-independent EEG results. Table 4 lists these informative musical features, which consistently appeared in above half of the subjects.

Related Figures (13)

FIGURE 1 | Electrode placements of 32 channels according to the international 10-20 system. For each of 16 30-s EEG trials, the short-time Fourier trans- form with non-overlapping 1-s Hamming window was applied
FIGURE 1 | Electrode placements of 32 channels according to the international 10-20 system. For each of 16 30-s EEG trials, the short-time Fourier trans- form with non-overlapping 1-s Hamming window was applied
Table 1 | A summary of EEG feature types.
Table 1 | A summary of EEG feature types.
FIGURE 2 | The valence and arousal classification results using the subject-dependent EEG feature sets with/without the F-score basec feature selection. The numbers above the bars represent the mean values of the results, whereas the numbers in bold indicate the accuracies significantly better (jp < 0.01) than the majority voting accuracy (valence: ~63%, arousal: ~61%). ‘Indicates that the accuracy with feature selection significantly outperformed that without feature selection (p < 0.01).
FIGURE 2 | The valence and arousal classification results using the subject-dependent EEG feature sets with/without the F-score basec feature selection. The numbers above the bars represent the mean values of the results, whereas the numbers in bold indicate the accuracies significantly better (jp < 0.01) than the majority voting accuracy (valence: ~63%, arousal: ~61%). ‘Indicates that the accuracy with feature selection significantly outperformed that without feature selection (p < 0.01).
FIGURE 3 | The valence and arousal classification results using the subject-dependent multimodal approach with/without feature selection. The results of the subject-dependent EEG modality (feature type: MESH) and the music modality (feature type: MUSIC) are also provided for comparison. The numbers above the bars represent the mean values of the results, whereas the numbers in bold indicate the accuracies significantly better (jp < 0.01) than the majority voting accuracy (valence: ~63%, arousal: ~61%). ‘Indicates that the accuracy with feature selection significantly outperformed that without feature selection (p < 0.01).
FIGURE 3 | The valence and arousal classification results using the subject-dependent multimodal approach with/without feature selection. The results of the subject-dependent EEG modality (feature type: MESH) and the music modality (feature type: MUSIC) are also provided for comparison. The numbers above the bars represent the mean values of the results, whereas the numbers in bold indicate the accuracies significantly better (jp < 0.01) than the majority voting accuracy (valence: ~63%, arousal: ~61%). ‘Indicates that the accuracy with feature selection significantly outperformed that without feature selection (p < 0.01).
-IGURE 4 | The percent composition of contributions of EEG (DLAT, DCAU, and PSD) and musical (Pitch, Dissonance, Loudness, and MFCC) features o the subject-dependent multimodality. The composition of the subject-dependent EEG modality is also provided for comparison.
-IGURE 4 | The percent composition of contributions of EEG (DLAT, DCAU, and PSD) and musical (Pitch, Dissonance, Loudness, and MFCC) features o the subject-dependent multimodality. The composition of the subject-dependent EEG modality is also provided for comparison.
the nodes represent the mean values of the results. tIndicates that the accuracy with feature selection significantly outperformed that without feature selection (p < 0.01), yet were comparable (p > 0.1) to majority voting accuracies (valence: ~63%, arousal: ~61%). FIGURE 5 | The valence and arousal classification results of the subject-independent EEG features (type: MESH) in term of the average number of features, electrodes, and accuracies using with/without feature selection under the LFI criteria (0.1 ~ 0.6). The numbers near to
the nodes represent the mean values of the results. tIndicates that the accuracy with feature selection significantly outperformed that without feature selection (p < 0.01), yet were comparable (p > 0.1) to majority voting accuracies (valence: ~63%, arousal: ~61%). FIGURE 5 | The valence and arousal classification results of the subject-independent EEG features (type: MESH) in term of the average number of features, electrodes, and accuracies using with/without feature selection under the LFI criteria (0.1 ~ 0.6). The numbers near to
FIGURE 6 | The topographic mapping of informative EEG features consistently appeared in multiple subjects. The rightmost topography colorcodes the importance of electrodes according to how frequent the electrodes were used to derive the corresponding features.
FIGURE 6 | The topographic mapping of informative EEG features consistently appeared in multiple subjects. The rightmost topography colorcodes the importance of electrodes according to how frequent the electrodes were used to derive the corresponding features.
mean values of the results, whereas the numbers in bold indicate the accuracies significantly better (jp < 0.02) than the majority voting accuracy (valence: ~63%, arousal: ~61%). Tindicates that the accuracy with feature selection significantly outperformed that without feature selection (p < 0.01). FIGURE 7 | The valence and arousal classification results using the subject-independent multimodal approach (LFI = 0.6) with/without feature selection. The results of the subject-independent EEG modality (feature type: MESH) and the music modality (feature type: MUSIC) are also provided for comparison. The numbers above the bars represent th
mean values of the results, whereas the numbers in bold indicate the accuracies significantly better (jp < 0.02) than the majority voting accuracy (valence: ~63%, arousal: ~61%). Tindicates that the accuracy with feature selection significantly outperformed that without feature selection (p < 0.01). FIGURE 7 | The valence and arousal classification results using the subject-independent multimodal approach (LFI = 0.6) with/without feature selection. The results of the subject-independent EEG modality (feature type: MESH) and the music modality (feature type: MUSIC) are also provided for comparison. The numbers above the bars represent th
FIGURE 8 | The percent composition of contributions of EEG (DLAT, DCAU, and PSD) and musical (Pitch, Dissonance, Loudness, and MFCC) features to the subject-independent multimodality. The composition of the subject-independent EEG modality is also provided for comparison.
FIGURE 8 | The percent composition of contributions of EEG (DLAT, DCAU, and PSD) and musical (Pitch, Dissonance, Loudness, and MFCC) features to the subject-independent multimodality. The composition of the subject-independent EEG modality is also provided for comparison.
Table 4 | The informative musical features in the subject-independent multimodal approach.
Table 4 | The informative musical features in the subject-independent multimodal approach.
Related topics:
BioinformaticsCognitive ScienceMusic Emotion Classification
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved