RONI_defence - abstractdetect_poison not working

[RaMAT-44]

Hi Team,

Working on a use case to detect poison using RONI_defence on the adversarial data after RONIDefense is initialized called arbitrary method detect_poison. It returns an error on

before_classifier.fit(x_suspect,y_suspect)

[beat-buesser]

Hi @TRKSasank Thank you very much for using ART! How can we reproduce the error you are observing?