Stochastic Grad Descent for nn

sgd.R

 SGD <- function(training_data, epochs, mini_batch_size, lr, C, sizes, num_layers, biases, weights,
                 verbose=FALSE, validation_data)
 {
   # Every epoch
   for (j in 1:epochs){
     # Stochastic mini-batch (shuffle data)
     training_data <- sample(training_data)
     # Partition set into mini-batches
     mini_batches <- split(training_data, 
                           ceiling(seq_along(training_data)/mini_batch_size))
     # Feed forward (and back) all mini-batches
     for (k in 1:length(mini_batches)) {
       # Update biases and weights
       res <- update_mini_batch(mini_batches[[k]], lr, C, sizes, num_layers, biases, weights)
       biases <- res[[1]]
       weights <- res[[-1]]
     }
   }
   # Return trained biases and weights
   list(biases, weights)
 }