An ML engineer at an entertainment company is refining an ML model to predict audience preferences. As the model learns from data, the ML engineer notices that the model’s performance peaks early and then begins to gradually decline. The ML engineer must prevent the performance degradation after initial success.
Which solution will meet this requirement?
A. Increase the number of layers.
B. Implement early stopping.
C. Add more neurons to each layer to capture complex patterns.
D. Examine model bias and variance to understand performance issues.
Show Answer
Correct Answer: B
Explanation: A model whose validation performance improves initially and then degrades with continued training is exhibiting overfitting. Early stopping halts training when validation performance stops improving, preventing further overfitting. Increasing layers or neurons typically increases model capacity and can worsen overfitting, while examining bias and variance is useful for diagnosis but does not itself prevent the degradation.
Question 22
An ML engineer decides to use Amazon SageMaker AI automated model tuning (AMT) for hyperparameter optimization (HPO). The ML engineer requires a tuning strategy that uses regression to slowly and sequentially select the next set of hyperparameters based on previous runs. The strategy must work across small hyperparameter ranges.
Which solution will meet these requirements?
A. Grid search
B. Random search
C. Bayesian optimization
D. Hyperband
Show Answer
Correct Answer: C
Explanation: Bayesian optimization builds a surrogate regression model of the objective function from previous trials and sequentially chooses the next hyperparameter configuration by balancing exploration and exploitation. This is the SageMaker AMT strategy that uses prior runs and is effective for relatively small hyperparameter search spaces.
Question 23
An ML engineer needs to build a processing pipeline to identify and remove personally identifiable information (PII) from petabytes of unstructured data. The ML engineer will use the processed data to train ML models in Amazon SageMaker AI.
Which solution will meet these requirements?
A. Use the Apache Spark-based serverless engine from AWS Glue interactive sessions. Use the Detect PII transform feature to identify and remove the PII data.
B. Use AWS Glue Data Wrangler within Amazon SageMaker Canvas to detect and remove the PII.
C. Use the Amazon SageMaker Clarify API to detect and mask the PII data.
D. Use the DetectEntities API action in Amazon Comprehend to identify and remove the PII data.
Show Answer
Correct Answer: A
Explanation: AWS Glue provides a scalable Spark-based serverless data processing engine with a built-in Detect PII transform designed to identify and redact or remove PII from large-scale datasets. It is appropriate for processing petabytes of unstructured data before training Amazon SageMaker models. Data Wrangler in SageMaker Canvas is not intended for petabyte-scale pipelines, SageMaker Clarify focuses on bias and explainability rather than PII detection, and Amazon Comprehend DetectEntities is not the dedicated large-scale Glue PII transformation for this use case.
Question 24
A healthcare company wants to detect irregularities in patient vital signs that could indicate early signs of a medical condition. The company has an unlabeled dataset that includes patient health records, medication history, and lifestyle changes.
Which algorithm and hyperparameter should the company use to meet this requirement?
A. Use the Amazon SageMaker AI XGBoost algorithm. Set max_depth to greater than 100 to regulate tree complexity.
B. Use the Amazon SageMaker AI k-means clustering algorithm. Set k to determine the number of clusters.
C. Use the Amazon SageMaker AI DeepAR algorithm. Set epochs to the number of training iterations.
D. Use the Amazon SageMaker AI Random Cut Forest (RCF) algorithm. Set num_trees to greater than 100.
Show Answer
Correct Answer: D
Explanation: The requirement is anomaly detection on an unlabeled dataset. Amazon SageMaker Random Cut Forest (RCF) is designed for unsupervised anomaly detection and identifies outliers in multidimensional data. The num_trees hyperparameter controls the size of the forest; using more than 100 trees can improve detection stability at the cost of additional compute. XGBoost is supervised, k-means performs clustering rather than anomaly detection, and DeepAR is for time-series forecasting.
Question 25
An ML engineer is developing a linear regression ML model. The model shows high accuracy on the training dataset but performs poorly on unseen new data.
Which action should the ML engineer take to address this issue?
A. Increase the complexity of the model to capture more patterns in the training data. Use Amazon SageMaker Debugger to monitor for convergence issues.
B. Apply ML techniques such as cross-validation and regularization. Use Amazon SageMaker Experiments to track and compare different model versions and their performance metrics.
C. Directly deploy the model into production. Use Amazon SageMaker Clarify to interpret model outputs on new data. Adjust the model based on these insights.
D. Increase the size of the training dataset without adjusting the size of the model. Retrain the model on the new data. Generate a confusion matrix to analyze the results.
Show Answer
Correct Answer: B
Explanation: The model fits the training data well but generalizes poorly to unseen data, which is the classic pattern of overfitting. Appropriate remedies include cross-validation to estimate generalization performance and regularization (such as L1/L2) to reduce overfitting. SageMaker Experiments is suitable for tracking and comparing model runs and performance. Increasing complexity would likely worsen overfitting, deploying before fixing the issue is inappropriate, and simply adding more data without other changes is not the best targeted action; a confusion matrix is also not applicable to linear regression.
Question 26
A company uses ML models to predict whether transactions are fraudulent. The company needs to identify as many fraudulent transactions as possible.
Which evaluation metric should the company use to evaluate the models to meet this requirement?
A. F1 score
B. Area Under the ROC Curve (AUC)
C. Precision
D. Recall
Show Answer
Correct Answer: D
Explanation: Recall measures the proportion of actual fraudulent transactions that are correctly identified (true positive rate). When the priority is to identify as many fraudulent transactions as possible, minimizing false negatives is the key objective, making recall the most appropriate evaluation metric.
Question 27
A company has developed a computer vision model. The company needs to deploy the model into production on Amazon SageMaker AI. The company has not hosted a model on SageMaker AI previously.
An ML engineer needs to implement a solution to track model versions. The solution also must provide recommendations about which Amazon EC2 instance types to use to host the model.
Which solution will meet these requirements?
A. Register the model in Amazon Elastic Container Registry (Amazon ECR). Use AWS Compute Optimizer for recommendations about instance types.
B. Register the model in the SageMaker Model Registry. Use SageMaker Autopilot for recommendations about instance types.
C. Register the model in the SageMaker Model Registry. Use SageMaker Inference Recommender for recommendations about instance types.
D. Register the model in Amazon Elastic Container Registry (Amazon ECR). Use SageMaker Experiments for recommendations about instance types.
Show Answer
Correct Answer: C
Explanation: SageMaker Model Registry is the managed feature for tracking and versioning ML models for deployment. SageMaker Inference Recommender analyzes model performance and provides recommendations for suitable SageMaker hosting instance types and configurations, especially useful for first-time deployments. Amazon ECR stores container images, not model versions, Compute Optimizer does not recommend SageMaker endpoint instance types, and Autopilot is for automated model building rather than inference instance selection.
Question 28
An ML engineer is using an Amazon SageMaker Studio notebook to train a neural network by creating an estimator. The estimator runs a Python training script that uses Distributed Data Parallel (DDP) on a single instance that has more than one GPU.
The ML engineer discovers that the training script is underutilizing GPU resources. The ML engineer must identify the point in the training script where resource utilization can be optimized.
Which solution will meet this requirement?
A. Use Amazon CloudWatch metrics to create a report that describes GPU utilization over time.
B. Add SageMaker Profiler annotations to the training script. Run the script and generate a report from the results.
C. Use AWS CloudTrail to create a report that describes GPU utilization and GPU memory utilization over time.
D. Create a default monitor in Amazon SageMaker Model Monitor and suggest a baseline. Generate a report based on the constraints and statistics the monitor generates.
Show Answer
Correct Answer: B
Explanation: SageMaker Profiler is designed to identify performance bottlenecks in training jobs at the code level. By adding SageMaker Profiler annotations to the training script, the engineer can collect detailed profiling data and generate reports that show where GPU resources are underutilized, including time spent in data loading, computation, and synchronization. CloudWatch provides high-level metrics only, CloudTrail is for API auditing rather than GPU profiling, and SageMaker Model Monitor is intended for monitoring deployed model/data quality, not training performance.
Question 29
A music streaming company constantly streams song ratings from an application to an Amazon S3 bucket. The company wants to use the ratings as an input for training and inference of an Amazon SageMaker AI model.
The company has an AWS Glue Data Catalog that is configured with the S3 bucket as the source. An ML engineer needs to implement a solution to create a repository for this data. The solution must ensure that the data stays synchronized during batch training and real-time inference.
Which solution will meet these requirements?
A. Ingest data into SageMaker Feature Store from the S3 bucket. Apply tags and indexes.
B. Use Amazon Athena. Create tables by using CREATE TABLE AS SELECT (CTAS) queries to group data.
C. Use AWS Lake Formation. Apply tag-based control on the data.
D. Use the Generate Data Insights function in SageMaker Data Wrangler.
Show Answer
Correct Answer: A
Explanation: Amazon SageMaker Feature Store is designed to serve as a centralized repository for ML features, supporting both an offline store for batch training and an online store for low-latency real-time inference. It integrates with data in Amazon S3 and the AWS Glue Data Catalog, and keeps the online and offline feature data synchronized. The other options provide analytics, governance, or data preparation capabilities rather than a synchronized ML feature repository.
Question 30
HOTSPOT
-
A company is using Amazon SageMaker to deploy a new version of its ML model.
Select the correct SageMaker traffic shifting strategy from the following list for each use case. Each traffic shifting strategy should be selected one time. (Choose three.)
• All at once traffic shifting
• Canary traffic shifting
• Linear traffic shifting
Show Answer
Correct Answer: Shift traffic in two steps to validate the new model version: Canary traffic shifting
Shift traffic to the new model version in the shortest amount of time: All at once traffic shifting
Incrementally shift traffic to the new model version over time: Linear traffic shifting
Explanation: Canary sends a small portion of traffic first, then the remainder after validation. All at once immediately routes all traffic to the new version. Linear gradually increases traffic in fixed increments over time.
$19
Get all 235 questions with detailed answers and explanations
Instant download HTML + PDF delivered the moment payment clears.
Secure Stripe checkout we never see or store your card details.
7-day refund if files are defective see our refund policy.
