Auxilliary Confidence Loss

An additional term added to a Loss Function that can make the model more confident in decisions over-time. A supervisor model can help decide this term (Weak to Strong Generalization)