http://d2l.ai/chapter_recurrent-neural-networks/sequence.html
Hi Doctor Li, will there be Tensorflow code posted for Chapter 8 to 17 later? Thank you very much!
Yes, we’re working on it.
Thank you very much! Appreciate it!
Hey guys! For exercise 1. for this section, how does writing a sine and cosine as a differential equation help us, analyze if more or less observations would help or not help with noise?
For exercise 1 part 1 dont we have to also increase the number of epochs with the number of observations? Or is the number of epochs a control variable in this experiment?
What would be a good way to set the features constant in the provided code, so that all of the plots and other code still works?
Hi @smizerex, great question! Since the sine and cosine functions are differentiable, we can calculate each point’s gradient, and then estimate the next potential point with the gradient times with Δx. But the estimate point should be with in the range of error 𝜖.
2 Likes
But we do not know that it is a sin function in the first place.