3 edited tags
| link
2 added 49 characters in body
source | link

Data augmentation - training Training a model on samples with both correct and wrong (non-noise) labels

When we train a model, we usually use pairs of $(x_n,\ label_n)$ as training data. In my case, I have pairs of $(x_n, \ label_n)$ and also pairs of $(x_n,\ wrong\_label_n)$, where both $label$ and $wrong\_label$ come from the same set of labels. Note: the wrong labels are known, not noise.

Is there any way to train on samples with known wrong and correct labels instead of training on correct labels only? Maybe a modified loss function I can use? I don't really mind what kind of model it works for (regressions, CART, ANNs etc).

Data augmentation - training a model on samples with both correct and wrong labels

When we train a model, we usually use pairs of $(x_n,\ label_n)$ as training data. In my case, I have pairs of $(x_n, \ label_n)$ and also pairs of $(x_n,\ wrong\_label_n)$, where both $label$ and $wrong\_label$ come from the same set of labels.

Is there any way to train on samples with known wrong and correct labels instead of training on correct labels only? Maybe a modified loss function I can use? I don't really mind what kind of model it works for (regressions, CART, ANNs etc).

Training a model on samples with both correct and wrong (non-noise) labels

When we train a model, we usually use pairs of $(x_n,\ label_n)$ as training data. In my case, I have pairs of $(x_n, \ label_n)$ and also pairs of $(x_n,\ wrong\_label_n)$, where both $label$ and $wrong\_label$ come from the same set of labels. Note: the wrong labels are known, not noise.

Is there any way to train on samples with known wrong and correct labels instead of training on correct labels only? Maybe a modified loss function I can use? I don't really mind what kind of model it works for (regressions, CART, ANNs etc).

1
source | link

Data augmentation - training a model on samples with both correct and wrong labels

When we train a model, we usually use pairs of $(x_n,\ label_n)$ as training data. In my case, I have pairs of $(x_n, \ label_n)$ and also pairs of $(x_n,\ wrong\_label_n)$, where both $label$ and $wrong\_label$ come from the same set of labels.

Is there any way to train on samples with known wrong and correct labels instead of training on correct labels only? Maybe a modified loss function I can use? I don't really mind what kind of model it works for (regressions, CART, ANNs etc).