Fig. 23

Precision-recall curves of LSTM-AE+NN model [32] with three different batch sizes (32, 256 and 1024) on three datasets of increasing size. The higher the curve, the better performance with a horizontal dashed line to represent random prediction. An ideal precision-recall curve would cross the (1,1) point. The splice data has three panels since precision-recall curves assume binary classification and the splice dataset has three classes (0, 1, 2). Each panel corresponds to prediction one class vs the other two combined. Unlike other models, there is a clear distinction in the class 0 prediction performance of this model compared to other classes in the splice data. It appears that class 0 is harder to predict with a lower recall for a given precision value compared to the other classes