I would like to create a RNN, that uses one (A) or several time series (with the same length, A, B, C...) as an input and creates another time series (Z) as an output from that . Basically all time series are derived from on time series, so their relationship that can be expressed mathematically. ex. B = dA/dt, ... C = moving mean(B)

For training I want to feed the NN with the input and output time series, which I can create with a simulation with known parameters, so that I know the output is correct.

I am new to working with neural networks and couldn't find a NN that seemed to do a similar job. Does somebody have a NN in mind that could do the job?

I asked chatGPT, which said that simple RNNs with or without LSTM or GRU cells could do the job.

Any ideas?



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.