How to use parallel computing for a custom optimization function in matlab?

Currently I am trying to train a neural network of following configuration

• Input Layer size -> 307200
• Hidden Layer size -> 50
• Output Layer size -> 10

I use the optimization function provided by coursera ' fmincg' to train the network. Since the network size is big, its taking a long time. So I tried to use parallel computing toolbox in matlab for this. Matlab has a documentation for using parallel computing for optimization functions provided by matlab(fminunc etc).

But I don't know how to use parallel computing for a custom optimization function for a user defined function.

If someone can throw some light on this, I would be grateful.