1

I'm reading this post that describes how to train LSTMs with variable time step lengths. But does that have repercussions? Should I preprocess the time series in to varying permutations? e.g. Should the input user_a,bad,[(t1,req1), (t2,req2), (t3,req3)] be divided into something like:

user_a,bad,[(t1,req1)]
user_a,bad,[(t1,req1), (t2,req2)]
user_a,bad,[(t1,req1), (t2,req2), (t3,req3)]

For context, I'm looking to classify bad users based on their requests. All requests from bad users are bad, but presumably, more requests gives the model more signals for evaluations. But at prediction time, they come in one at a time.

Or some other ways to preprocess the time-series:

user_a,bad,[(t1,req1)]
user_a,bad,[(t2,req2), (t3,req3)]
user_a,bad,[(t1,req1), (t3,req3)]
Ethan
  • 1,633
  • 9
  • 24
  • 39
bli00
  • 111
  • 1
  • I am not sure what exactly you are asking, but the question and the answer to the post you mentioned says that in order to use variable length sequences in LSTM in keras you have to either pad them to max length or to do some extra work by grouping your training data into batches according to the length of the input sequence. Also your first suggested option of representing the input looks better to me. – Valentas Dec 17 '21 at 10:51

0 Answers0