
Answer-first summary for fast verification
Answer: Decrease the recall to make the model more conservative in its predictions, reducing false positives.
Decreasing the recall by adjusting the softmax threshold makes the model more conservative in its predictions. This approach reduces the number of false positives, thereby increasing precision. It is particularly effective when precision is a priority over recall, aligning with the business requirements for high accuracy in image categorization.
Author: LeetQuiz Editorial Team
Ultimate access to all questions.
In your role at a leading social media company, you are responsible for developing a machine learning model to automatically identify and categorize images that contain cars. The model uses an object detection neural network, and each training example is associated with a single class. After deploying the model to AI Platform Prediction for evaluation, you observe that the precision metrics do not meet the stringent business requirements set by your company. The current model's performance is critical for ensuring user satisfaction and compliance with data accuracy standards. Given the scenario, which of the following adjustments to the model's softmax threshold in the final layer would most effectively improve precision, considering the trade-offs between precision and recall? Choose the best option.
A
Decrease the false negative rate to ensure more positive predictions are correct.
B
Increase the recall to capture more true positives, even if it means more false positives.
C
Decrease the recall to make the model more conservative in its predictions, reducing false positives.
D
Increase the false positive rate to allow more leniency in predictions, potentially capturing more true positives.
No comments yet.