Hello Akinyemi Moruff Oyelakin

I am not sure if solution exist in the world for this. Noisy or ambiguous data is a bottle neck in any decision making problem. Example if I am teaching a child to classify cat and dog, today I showed him a cat and tell this cat. Next day I will tell him its dog. Basically he can't learn anything.

So if the data is ambiguous then

1) Use noise cleansing technique

2) Learning more feature of the data, which may reduce the ambiguity up to some level. Like a cat and a tiger may looks similar in gray scale. but if you add RGB feature the ambiguity will be gone.

All the best

Dear Akinyemi Moruff Oyelakin ,

Look the link, maybe useful.

Article Ensemble Feature Selection: Homogeneous and Heterogeneous Approaches

Regards,

Shafagat