Research Article

DQfD-AIPT: An Intelligent Penetration Testing Framework Incorporating Expert Demonstration Data

Table 5

Hyperparameter setting of the algorithm.

HyperparameterDQfDDQN

Batch size512512
Epsilon0.90.9
Discount factor0.0150.015
Epsilon exponential decay50005000
Epsilon minimum0.10.1
Learning rate0.010.01
Demonstration memory size1000
Replay memory size50005000
Target network update frequency66
Max steps per episode30003000
Training episode200200