The initial trainset is likely to contain a lot of noise, so filtering based on this data may be misleading since the models in the ensemble will be affected by this noise.
To address that. We first remove most of the potential noisy examples. The output will be a noise-free set that will be used in the Noise-free filtering step.
salvacorts
commented
3 years ago
The initial trainset is likely to contain a lot of noise, so filtering based on this data may be misleading since the models in the ensemble will be affected by this noise.
To address that. We first remove most of the potential noisy examples. The output will be a noise-free set that will be used in the Noise-free filtering step.