EOG feature relevance determination for microsleep detection
Automatic relevance determination (ARD) was applied to two-channel EOG recordings for microsleep event (MSE) recognition. 10 s immediately before MSE and also before counterexamples of fatigued, but attentive driving were analysed. Two type of signal features were extracted: the maximum cross correl...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
De Gruyter
2017-09-01
|
| Series: | Current Directions in Biomedical Engineering |
| Subjects: | |
| Online Access: | https://doi.org/10.1515/cdbme-2017-0053 |
| _version_ | 1797848750532591616 |
|---|---|
| author | Golz Martin Wollner Sebastian Sommer David Schnieder Sebastian |
| author_facet | Golz Martin Wollner Sebastian Sommer David Schnieder Sebastian |
| author_sort | Golz Martin |
| collection | DOAJ |
| description | Automatic relevance determination (ARD) was applied to two-channel EOG recordings for microsleep event (MSE) recognition. 10 s immediately before MSE and also before counterexamples of fatigued, but attentive driving were analysed. Two type of signal features were extracted: the maximum cross correlation (MaxCC) and logarithmic power spectral densities (PSD) averaged in spectral bands of 0.5 Hz width ranging between 0 and 8 Hz. Generalised learn-ing vector quantisation (GRLVQ) was used as ARD method to show the potential of feature reduction. This is compared to support-vector machines (SVM), in which the feature reduction plays a much smaller role. Cross validation yielded mean normalised relevancies of PSD features in the range of 1.6 – 4.9 % and 1.9 – 10.4 % for horizontal and vertical EOG, respectively. MaxCC relevancies were 0.002 – 0.006 % and 0.002 – 0.06 %, respectively. This shows that PSD features of vertical EOG are indispensable, whereas MaxCC can be neglected. Mean classification accuracies were estimated at 86.6±b 1.3 % and 92.3±b 0.2 % for GRLVQ and SVM, respectively. GRLVQ permits objective feature reduction by inclusion of all processing stages, but is not as accurate as SVM. |
| first_indexed | 2024-04-09T18:32:50Z |
| format | Article |
| id | doaj.art-3cb5760b1f254b379e6cc3f5d9863092 |
| institution | Directory Open Access Journal |
| issn | 2364-5504 |
| language | English |
| last_indexed | 2024-04-09T18:32:50Z |
| publishDate | 2017-09-01 |
| publisher | De Gruyter |
| record_format | Article |
| series | Current Directions in Biomedical Engineering |
| spelling | doaj.art-3cb5760b1f254b379e6cc3f5d98630922023-04-11T17:07:13ZengDe GruyterCurrent Directions in Biomedical Engineering2364-55042017-09-013226126410.1515/cdbme-2017-0053cdbme-2017-0053EOG feature relevance determination for microsleep detectionGolz Martin0Wollner Sebastian1Sommer David2Schnieder Sebastian3University of Applied Sciences Schmalkalden, Germany University of Applied Sciences Schmalkalden University of Applied Sciences Schmalkalden Institute of Experimental Psychophysio-logy GmbH, Düsseldorf, GermanyAutomatic relevance determination (ARD) was applied to two-channel EOG recordings for microsleep event (MSE) recognition. 10 s immediately before MSE and also before counterexamples of fatigued, but attentive driving were analysed. Two type of signal features were extracted: the maximum cross correlation (MaxCC) and logarithmic power spectral densities (PSD) averaged in spectral bands of 0.5 Hz width ranging between 0 and 8 Hz. Generalised learn-ing vector quantisation (GRLVQ) was used as ARD method to show the potential of feature reduction. This is compared to support-vector machines (SVM), in which the feature reduction plays a much smaller role. Cross validation yielded mean normalised relevancies of PSD features in the range of 1.6 – 4.9 % and 1.9 – 10.4 % for horizontal and vertical EOG, respectively. MaxCC relevancies were 0.002 – 0.006 % and 0.002 – 0.06 %, respectively. This shows that PSD features of vertical EOG are indispensable, whereas MaxCC can be neglected. Mean classification accuracies were estimated at 86.6±b 1.3 % and 92.3±b 0.2 % for GRLVQ and SVM, respectively. GRLVQ permits objective feature reduction by inclusion of all processing stages, but is not as accurate as SVM.https://doi.org/10.1515/cdbme-2017-0053automatic relevance determinationmicrosleepelectrooculographysupport-vector machines |
| spellingShingle | Golz Martin Wollner Sebastian Sommer David Schnieder Sebastian EOG feature relevance determination for microsleep detection Current Directions in Biomedical Engineering automatic relevance determination microsleep electrooculography support-vector machines |
| title | EOG feature relevance determination for microsleep detection |
| title_full | EOG feature relevance determination for microsleep detection |
| title_fullStr | EOG feature relevance determination for microsleep detection |
| title_full_unstemmed | EOG feature relevance determination for microsleep detection |
| title_short | EOG feature relevance determination for microsleep detection |
| title_sort | eog feature relevance determination for microsleep detection |
| topic | automatic relevance determination microsleep electrooculography support-vector machines |
| url | https://doi.org/10.1515/cdbme-2017-0053 |
| work_keys_str_mv | AT golzmartin eogfeaturerelevancedeterminationformicrosleepdetection AT wollnersebastian eogfeaturerelevancedeterminationformicrosleepdetection AT sommerdavid eogfeaturerelevancedeterminationformicrosleepdetection AT schniedersebastian eogfeaturerelevancedeterminationformicrosleepdetection |