problem understanding data input setup

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

problem understanding data input setup

BobM
This post was updated on .
Hi

I have used weka in the past - various classifiers - all programmed in java
- and all working satifactory :)

I am now delving into deep learning LSTM etc.

I wish to construct a recurrent neural network where

1) input instances are records consisting of 5 fields
a time stamp field
3 forex prices
and the final field is the classifier field (0, 1, 2) [representing "Up",
Down" and "No trade"]

My problem: I wish the input to the neural network to be a multiple number
of instances (say 20)
i.e. the network reads 20 consecutive instances and outputs 1 classification
(0, 1 or 2) [representing the trend of the next 20 instances]
i.e. the training is to be conducted on 20 instances each time - produce a
classification - and repeat by moving the window 1 instance forward, so 20
instances again for training consisting of 19 previous instances and 1 new
one

Can this be handled and if so, how does one code this in java?

I understand that one could construct each instance with 20 sets of inputs and a classification field but I am hoping one can leave the instances with just the 5 fields as outlined above

Bob M





--
Sent from: https://weka.8497.n7.nabble.com/
_______________________________________________
Wekalist mailing list -- wekalist@list.waikato.ac.nz
Send posts to wekalist@list.waikato.ac.nz
To unsubscribe send an email to wekalist-leave@list.waikato.ac.nz
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html