I’ve completed Part 1 of the Fast.AI course and have my bearings somewhat, but I’m trying to find the best way to represent/train etc. a model that takes a electrical current over time input and returns a number, indicating a number of events. I think this is called a regression output model, but I’m struggling on the input representation. I’ve tried creating an image of the current, but that didn’t work well. I’d like to use something tabular, but all the tabular examples I can find are more excel table based and don’t hold an entire (non-fixed size) array. I’d love something like
input = [[2,6,1,4,7],2] where [2,6,1,4,7] is the input array and 2 is the output.
1 Like
Hello,
Due to the nature of your task, I believe it would be more apt to employ sequence models such as recurrent neural networks (RNN) or transformers. It may also be worthwhile to take a look at traditional time series forecasting techniques.
1 Like