0

I would like to predict multiple timesteps into the future. My current NN outputs a sparse classification of 0, 1 or 2.

Sparse classifications outputs via a SoftMax Dense layer with 3 neurons to correspond to the three categories mentioned above.

How would I shape the output layer (softmaxed Dense) to give me the ability to predict two timesteps into the future, while keeping the sparse categorical classes to only 3?

Right now if I set that Dense layer to have 6 neurons (3 classes * 2 timesteps) I get an output of a sparse categorical classification with 6 classes and 1 timestep.

Book Of Zeus
  • 49,509
  • 18
  • 174
  • 171
Just_Learning
  • 31
  • 1
  • 5
  • Please provide some code to show some effort. – Benjamin Breton May 20 '19 at 06:52
  • It's more of a theoretical question. Let me rephrase is for clarity: How would one set up, in a theoretical sparse categorical temporal NN, a situation where more than one timesteps are predicted into the future given the fact that the output Dense layer only outputs for the number of categories (as defined by the 'units' parameter in the Keras documentation) - and any addition to that by increasing the number of units causes that Dense layer to create new bogus categories (for example changing the actual number of categories from 3 to 6 as per the math outlined in the question) – Just_Learning May 20 '19 at 07:58

0 Answers0