FXstreet

We revisit a form of neural network we had considered in an earlier article by dwelling on one specific hyperparameter. The learning-rate. The Generative Adversarial Network is a neural network that operates in pairs, where one network is trained traditionally to discern the truth, while another is trained to discern the former’s projections from real occurrences. This duality does imply that the traditionally trained network (the former) is trying to fool the latter and this is true, however

...

Let's get acquainted with a new model family: Ordinary Differential Equations. Instead of specifying a discrete sequence of hidden layers, they parameterize the derivative of the hidden state using a neural network. The results of the model are calculated using a "black box", that is, the Differential Equation Solver. These continuous-depth models use a constant amount of memory and adapt their estimation strategy to each input signal. Such models were first introduced in the paper "Neural Ordinary Differential Equations".

...