Why does my time series output of my NARXNET have less elements than my input?

Hi all,

I trained a neural network using narxnet and generated the Simulink block of it using gensim. When I delivered a time series into the input port of the neural network block (and fed back the output to the feedback signal port), I obtained a time series as the output with a length that was less than my input size. My original input time series size is 60001x1, but my output is only 61x1. Why does this happen?

Thank you,

Ghazi

Answer 1

[Greg Heath]

Because the first few data points are used as delays.

Hope this helps

Thank you for formally accepting my answer

Greg

[Ghazi Binarandi]

I think that's not the case. I just figured out that the sampling time was the reason. I didn't set the sampling time in 'gensim.' So the sampling time is set to be 1, the default value. Then, I copied the Simulink neural network block to the other Simulink file that has sampling time of 1/1000. I think that's why the element size is reduced from 60001 to 61. Thanks for the answer anyway!

[Greg Heath]

Sorry, I didn't read your whole question before I jumped to conclusions and posted the answer.

Greg the Embarrassed