Figure 3.

Statistically significant difference analysis of ten-fold cross-validation between our model and 1D-2C-CNN method. $\mathrm{Model}\_1, \mathrm{model}\_2,\dots , \mathrm{model}\_10$ are the models of CapsNet + CBAM by the ten-fold cross-validation. When p-value is less than 0.05, the difference between two models is significant. Since the p-values calculated by ten-fold cross-validation models are quite different, we provide the result of $-lo{g}_{10}$ (p-value) instead of p-value in this figure. When the p-value is higher than the baseline (p-value = 0.05), it indicates that the ten-fold cross-validation of our model is significantly different from the 1D-2C-CNN method.