leave_one_out_all : contains the split from the msad repo to reproduce the experiment. 
In the MSAD repo this split is used by splitting the train set into a train and validation 70/30.