Academia.eduAcademia.edu

Figure 2 – uploaded by sugondo hadiyoso

See full PDFdownloadDownload figure
Table 4 SVM kernel performance for 4th scenario Table 5 SVM kernel performance for 5th scenario

Table 4 SVM kernel performance for 4th scenario Table 5 SVM kernel performance for 5th scenario

Related Figures (5)

SVM kernel can be grouped into two, and they are linear SVM and nonlinear SVM. The nonlinear kernel consists of two polynomial functions (quadratic and cubic), and the radial basis function (fine, medium, and coarse gaussian). In this research, we use these six kernels to find out which one suitable for the classification process [24]. The dataset consists from 210 signals, that split into 120 training data and 90 evaluation data. Those data consist from 20 GNSZ, 50 FNSZ, 25 TCSZ, and 25 Normal EEG signals. On the other hand, the number of evaluation data was 90 signals that consist of 20 GNSZ, 50 FNSZ, 5 TCSZ, and 15 Normal EEG signals.
SVM kernel can be grouped into two, and they are linear SVM and nonlinear SVM. The nonlinear kernel consists of two polynomial functions (quadratic and cubic), and the radial basis function (fine, medium, and coarse gaussian). In this research, we use these six kernels to find out which one suitable for the classification process [24]. The dataset consists from 210 signals, that split into 120 training data and 90 evaluation data. Those data consist from 20 GNSZ, 50 FNSZ, 25 TCSZ, and 25 Normal EEG signals. On the other hand, the number of evaluation data was 90 signals that consist of 20 GNSZ, 50 FNSZ, 5 TCSZ, and 15 Normal EEG signals.
True positive (TP) is the correctly classified positive class epochs. True negative (TN) defines as the correctly classified negative class epochs. While the number of epochs that belong to the negative class but have been identified mistakenly as positive class, and the number of epochs that belong to the positive class but have been identified mistakenly as the negative class is the definition of false positive (FP), and false negative (FN) respectively [26]. Figure 2 Seizure type signal, GNSZ, FNSZ, TCSZ and normal EEG signal Table 1 SVM kernel performance for Ist scenario
True positive (TP) is the correctly classified positive class epochs. True negative (TN) defines as the correctly classified negative class epochs. While the number of epochs that belong to the negative class but have been identified mistakenly as positive class, and the number of epochs that belong to the positive class but have been identified mistakenly as the negative class is the definition of false positive (FP), and false negative (FN) respectively [26]. Figure 2 Seizure type signal, GNSZ, FNSZ, TCSZ and normal EEG signal Table 1 SVM kernel performance for Ist scenario
Table 2 SVM kernel performance for 2nd scenario Table 3 SVM kernel performance for 3rd scenario
Table 2 SVM kernel performance for 2nd scenario Table 3 SVM kernel performance for 3rd scenario
Je Se VHUDDEPLCULLOUTE INOUE We used R2018b version of Matlab software. We use MFCC toolbox for extracting 14 cepstra coefficients, and SVM toolbox for the classifier. We used 5-fold cross validation to test generalization capabilities. After that, we split the dataset into training and testing data as mentioned in section 2. Figure 2 shows the example of normalized seizure type signals, GNSZ, FNSZ, TCSZ, and norma EEG signal. Visually the pattern of each seizure signal shows that they follow the general information about the seizure, which has a lower frequency, and higher amplitude [2]. We nahcerve the antimim clacaficatian reailt fram different SVM kernele The kernele need
Je Se VHUDDEPLCULLOUTE INOUE We used R2018b version of Matlab software. We use MFCC toolbox for extracting 14 cepstra coefficients, and SVM toolbox for the classifier. We used 5-fold cross validation to test generalization capabilities. After that, we split the dataset into training and testing data as mentioned in section 2. Figure 2 shows the example of normalized seizure type signals, GNSZ, FNSZ, TCSZ, and norma EEG signal. Visually the pattern of each seizure signal shows that they follow the general information about the seizure, which has a lower frequency, and higher amplitude [2]. We nahcerve the antimim clacaficatian reailt fram different SVM kernele The kernele need
Related topics:
EngineeringComputer ScienceArtificial IntelligencePhysics
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved