A Deep Random Forest Model on Spark for Network Intrusion Detection

Figure 4

The architecture of deep PRF, the number of layers, is automatically determined based on the calculated performance gain, and the training process will be terminated if improvement is lower than the threshold.