Defining custom deep learning layer

I have been trying out the deep learning functionality within matlab for a project. I have two target classes '1' and '0'. I want to perform the following. I would like to penalize the misclassification of a '0' class to '1' higher than vice-versa. In SVMs, Treebaggers etc, I used to define a custom cost function. But I am unable to find that option within the deep learning functionality. Is there a fix for it or can you guide me to the appropriate work around?