MLA-C01 · Question #230
An ML engineer is tuning an image classification model that shows poor performance on one of two available classes during prediction. Analysis reveals that the images whose class the model performed…
The correct answer is B. Optimize for F1 score. Use image augmentation on the less common images to generate new. With a highly underrepresented class, accuracy can remain high even if the model performs poorly on that minority class. Optimizing for F1 better balances precision and recall for the minority class, and applying image augmentation increases the effective number and diversity…
Question
An ML engineer is tuning an image classification model that shows poor performance on one of two available classes during prediction. Analysis reveals that the images whose class the model performed poorly on represent an extremely small fraction of the whole training dataset. The ML engineer must improve the model’s performance. Which solution will meet this requirement?
Options
- AOptimize for accuracy. Use image augmentation on the less common images to generate new
- BOptimize for F1 score. Use image augmentation on the less common images to generate new
- COptimize for accuracy. Use Synthetic Minority Oversampling Technique (SMOTE) on the less
- DOptimize for F1 score. Use Synthetic Minority Oversampling Technique (SMOTE) on the less
How the community answered
(38 responses)- A3% (1)
- B68% (26)
- C11% (4)
- D18% (7)
Explanation
With a highly underrepresented class, accuracy can remain high even if the model performs poorly on that minority class. Optimizing for F1 better balances precision and recall for the minority class, and applying image augmentation increases the effective number and diversity of minority-class examples in a way that is appropriate for images.
Topics
Community Discussion
No community discussion yet for this question.