How to train too large data set using neural network

I'm using fitnet neural network in making regression for input data contains (2721) samples leads to one output all over the day (1440 m), I'm facing the problem of out of memory and if I divided the training process to subsets it start a new training process each time and lose the previous training

Answer 1

[Greg Heath]

Insufficient explanation. Need details,

 [ I N ] = size(input)  % [ ? 2721 ]
 [ O n ] = size(target) % [ ? 2721 ]

If you use subsets, design a separate net for each subset

Then design a final net with the subset outputs as input.

Hope this helps

Thank you for formally accepting my answer

Greg