Does TMVA MLP work if IgnoreNegWeightsInTraining = False

Hi experts,

I plan to train a neural net using a TMVA method and need to define both positive and negative event by event weights to cancel out the effects of some events in the training sample which are not of interest to the classification problem but are not clearly separated from the events we are interested in.

Is this possible with the TMVA MLP method if I set IgnoreNegWeightsInTraining = False?
According to the TMVA users guide, no neural network supports negative event by event weight.

Thanks in advance.