Q. What does 'model drift' refer to in the context of deployed models?
A.
The process of updating the model with new data
B.
The degradation of model performance over time due to changes in data distribution
C.
The initial training phase of the model
D.
The difference between training and testing datasets
Show solution
Solution
Model drift occurs when the statistical properties of the target variable change over time, leading to decreased model performance.
Correct Answer:
B
— The degradation of model performance over time due to changes in data distribution
Learn More →
Q. What is a common method for monitoring a deployed machine learning model?
A.
Cross-validation
B.
A/B testing
C.
Grid search
D.
K-fold validation
Show solution
Solution
A/B testing is commonly used to compare the performance of the deployed model against a baseline or alternative model in real-time.
Correct Answer:
B
— A/B testing
Learn More →
Q. What is a microservice architecture in the context of model deployment?
A.
A single monolithic application
B.
A method to deploy models on mobile devices
C.
A way to break down applications into smaller, independent services
D.
A technique for batch processing of data
Show solution
Solution
Microservice architecture allows applications to be divided into smaller, independent services, which can be developed, deployed, and scaled independently.
Correct Answer:
C
— A way to break down applications into smaller, independent services
Learn More →
Q. What is the role of a load balancer in model deployment?
A.
To train multiple models simultaneously
B.
To distribute incoming requests across multiple instances of a model
C.
To store model artifacts
D.
To preprocess input data
Show solution
Solution
A load balancer distributes incoming requests across multiple instances of a model to ensure efficient resource utilization and high availability.
Correct Answer:
B
— To distribute incoming requests across multiple instances of a model
Learn More →
Q. What is the significance of 'feature store' in model deployment?
A.
To store raw model outputs
B.
To manage and serve features for model training and inference
C.
To visualize feature importance
D.
To automate model retraining
Show solution
Solution
A feature store centralizes the management of features, making them easily accessible for both training and inference.
Correct Answer:
B
— To manage and serve features for model training and inference
Learn More →
Q. What is the significance of 'latency' in model deployment?
A.
It measures the model's accuracy
B.
It indicates the time taken to make predictions
C.
It refers to the amount of data processed
D.
It assesses the model's complexity
Show solution
Solution
Latency refers to the time taken for the model to process input data and return predictions, which is critical for user experience.
Correct Answer:
B
— It indicates the time taken to make predictions
Learn More →
Q. Which deployment strategy involves gradually rolling out a new model to a subset of users?
A.
Blue-green deployment
B.
Canary deployment
C.
Rolling deployment
D.
Shadow deployment
Show solution
Solution
Canary deployment gradually rolls out a new model to a small subset of users to monitor its performance before a full rollout.
Correct Answer:
B
— Canary deployment
Learn More →
Q. Which of the following is NOT a common evaluation metric for deployed models?
A.
Accuracy
B.
Precision
C.
Recall
D.
Training loss
Show solution
Solution
Training loss is not used for evaluating deployed models; it is a measure of performance during training, not in production.
Correct Answer:
D
— Training loss
Learn More →
Q. Which of the following is NOT a common method for monitoring deployed models?
A.
Performance metrics tracking
B.
User feedback collection
C.
Data versioning
D.
Real-time prediction logging
Show solution
Solution
Data versioning is not a monitoring method; it is a practice used to manage changes in datasets over time.
Correct Answer:
C
— Data versioning
Learn More →
Showing 1 to 9 of 9 (1 Pages)
