Noise-free filtering

[salvacorts]

An ensemble classifier is fitted with the resulting trainset of the preliminary filtering.

Then this ensemble is used to classify the whole set (including the example removed on the preliminary filtering step).

The output will be two sets:

• Clean set: Examples considered clean
• Noisy set: Examples considered as potentially noisy