What is the purpose of class weighting when training on imbalanced datasets?

Machine Learning Medium

Machine Learning — Medium

What is the purpose of class weighting when training on imbalanced datasets?

Key points

  • Class weighting adjusts loss contribution based on class frequency
  • Helps the model focus on minority class examples
  • Prevents bias towards majority class
  • Does not involve duplicating minority class examples in the batch

Ready to go further?

Start full exam Practice more questions

Related questions