Machine Learning-Based Stealing Attack of the Temperature Monitoring System for the Energy Internet of Things

Figure 2

The figure shows how to steal a model from a trained network. The entire network can be divided into two parts; that is, copycat is divided into two parts. First of all, we see that the part on the left is the existing network, which is what we call the target network. It uses the data we already have for training and testing, while the network on the right is the copycat network. Its basics are that it is based on a certain cognition of the target network, and its training set and the training set used by the original network are in the same problem domain.