Friday, September 25, 2020

Decision Making in Recurrent Neural Networks

 Decision Making in Recurrent Neural Networks



2 comments:

  1. Great article describing the feedback loop. Couldn't we get a stackoverflowexception because of the infinite recursion? Or do we only have the data set backwards until t-n where n=buffer size?

    ReplyDelete
  2. I don't think its like a memory bound application where stackoverflow and bufferoverflow type of exceptions due to
    memory related issues come over in Memory bound applications . Here we first create a neuron and then try to map our dataset whch obviously is a huge one ,and if this wouldnt have been read by pandas , scipy,numpy and other packages then ,then and there this sort of buffer and stack issue would have arose . Please note that this is already being handled by libraries designed to handle multitudes of data and create a neuron from those interpreted data , so buffer bound issues out of question if it wouldnt have been read before the creation of neuron

    ReplyDelete