A doubt in Deep Q Network Algorithm

Dear Administrator,

Could you please guide me on this issue?

According to the slide,

May i know the reason of training a new neural network (Label 2 in the figure) but we have already started training a neural network already (Label 1 in the figure)? What is the purpose of training a neural network (Label 1 in the figure) in this case?

Thank you

you are not training the neural network in label 2, you are looking for the value of Qnew in order to find the optimal action to take in a current state