Loss function in BackpropTrainer

[mfornet]

Is there any way to change the loss function when we are applying backpropagation to train the network? I want to use cross entropy (CE) instead of mean squared error (MSE) which is the default.

I have read this issue here, but I see no explicit answer, and in my opinion it is very important to have at least most common loss functions implemented. There is a way to do it, but it involves changing source code of pybrain.