I have a large data set for training neural network and computationally they can not be fed to NN because of computers lack of memory, So is it possible to train the NN with mini-batch by Levenberg-Marquardt algorithm?
Perhaps our paper " A convolutional neural network-based screening tool for X-ray serial crystallography" can shed some light to your problem of creating mini-batches.
I have read your paper, in the paper you used the mini batch with gradient descent algorithm, now what I want to know is that, is it possible to use mini batches in newtons method or Levenberg marquardt training?