Neural Networks

The approach to any physical problem begins with the elaboration of a theoretical model, which provides the reference framework on which said system can be studied and characterized, as well as making predictions that, through some experiment, are confirmed, validating the model, or are discarded, nullifying the validity of the model.
Recently, with the continuous improvement in calculation capabilities, machine learning techniques offer a very interesting alternative that dispenses with an underlying theoretical model, but is instead based on the training of neural networks from data obtained from the system.
This technique is of great interest in the prediction of future values of the time series studied and its applications have an impact on all branches of applied science.
In particular, our interest is based on the one of neural networks for testing random number generators and for use in chaos control schemes in lasers.