Answer-first summary for fast verification
Answer: Add an additional objective to penalize the model more for errors made on the minority class, and retrain the model., Upsample or reweight your existing training data, and retrain the model.
To address the issue of class imbalance and improve the model's performance on underrepresented subgroups, you should focus on strategies that enhance the model's sensitivity to the minority class. Option B ('Add an additional objective to penalize the model more for errors made on the minority class, and retrain the model') involves modifying the loss function to be more punitive towards errors on the minority class, which encourages the model to learn better from these examples. Option D ('Upsample or reweight your existing training data, and retrain the model') involves either duplicating the minority class samples or assigning higher weights to them, thereby increasing their representation during training. Both of these methods are effective in addressing class imbalance without the need for collecting additional data.
Author: LeetQuiz Editorial Team
Ultimate access to all questions.
You recently deployed an ML model in a production environment. After monitoring its performance for three months, you notice that the model is underperforming on certain subgroups, leading to biased predictions. You suspect that this issue arises due to class imbalances in the training data, but collecting additional data is not an option. Given these constraints, what actions should you take to address the model's inequitable performance? (Choose two.)
A
Remove training examples of high-performing subgroups, and retrain the model.
B
Add an additional objective to penalize the model more for errors made on the minority class, and retrain the model.
C
Remove the features that have the highest correlations with the majority class.
D
Upsample or reweight your existing training data, and retrain the model.
E
Redeploy the model, and provide a label explaining the model's behavior to users.
No comments yet.