Michele LaurelliBlog

Class imbalance

Class imbalance: A highly unequal class distribution that can make accuracy and training misleading. A classifier can achieve high accuracy while ignoring the minority class. Resampling, weighting, focal loss and metrics such as PR-AUC address the problem better than accuracy alone.

In italiano: Class ImbalanceLegge III · Come una rete impara dai dati

A classifier can achieve high accuracy while ignoring the minority class. Resampling, weighting, focal loss and metrics such as PR-AUC address the problem better than accuracy alone.

Examples

  • Fraud occurring below 1%
  • Rare manufacturing defects
  • Diagnosis of uncommon diseases