Google Professional Machine Learning Engineer Question 480
Single answerGoogle Cloud PlatformYou have deployed a machine learning model on Vertex AI Prediction for a financial risk assessment use case. To comply with regulatory requirements, you need to explain individual predictions made by the model to ensure transparency. Which approach should you take to achieve this in Vertex AI?
- A
Enable prediction logging in Vertex AI and analyze the logs for insights.
- B
Use the Explainable AI service in Vertex AI to generate feature attributions for the predictions.
- C
Deploy the model with a custom prediction container that implements manual feature attribution logic.
- D
Enable Data Labeling in Vertex AI to interpret the model predictions.
Show answer and explanation
Correct answer: B
Explanation
Vertex AI's Explainable AI service is specifically designed to provide feature attributions for model predictions. This capability enables you to understand the contribution of individual features to the prediction, which is necessary for compliance with regulatory requirements and ensuring model transparency. Other options either do not offer interpretability or are unrelated to the problem.
- A. Incorrect.
Prediction logging helps you monitor and debug the model but does not provide interpretability or feature attributions for individual predictions.
- B. Correct.
The Explainable AI service in Vertex AI provides feature attributions that explain the contribution of each feature to the model's prediction, meeting the requirement for transparency.
- C. Incorrect.
Implementing manual feature attribution logic in a custom container is unnecessary when Vertex AI already provides Explainable AI capabilities built into the platform.
- D. Incorrect.
Data Labeling in Vertex AI is used to create labeled datasets for training models and does not help with explaining predictions made by a deployed model.