Skip to main content
. 2018 May 22;18(5):1654. doi: 10.3390/s18051654

Figure 7.

Figure 7

A boxplot of the AUCs of different DNN architectures. For the LSTM architecture, at least two LSTM layers were involved, while for the ConvLSTM architecture, only one LSTM layer was involved. The dataset was split at (a) the subject level and (b) the sample level.