Is there a library for neural network reinforcement learning where the neural network is a black box?

Question

Neural networks

Good afternoon.
I am not very familiar with neural networks, but it seemed to me that it would be interesting to try using them to solve applied problems. For example, here we have the task of predicting the behavior of a forest fire.
Actually the question is: tell me the libraries for training the neural network for dummies. With a ready but not trained neural network implementation. Something like giving an array as input, getting an output, looking at the results and evaluating whether they are correct or not.
Still interested in the question, is it possible to organize reinforcement learning if the "game" includes several iterations of decision making, and evaluation of decisions will take place only at the end?
I don’t understand the question at all, and I apologize right away if I asked some kind of heresy :)

Answer 1

Neural networks are very different for different tasks. Look at keras as a tool, take a deep learning course on stepic or coursera as a preparation.