HOTSPOT
-
A company uses Amazon SageMaker AI to support ML workflows such as model training and deployment.
Select the correct registry from the following list to meet the requirements for each use case with the LEAST operational overhead. Each registry should be selected one or more times. (Choose four.)
Show Answer
Correct Answer: SageMaker Model Registry
Amazon Elastic Container Registry (Amazon ECR)
SageMaker Model Registry
Amazon Elastic Container Registry (Amazon ECR)
Explanation: Model Registry manages model packages, versions, tags, and model package groups for discoverability. Amazon ECR stores and distributes container images, including predefined/built-in images and custom images with language packages, kernels, and dependencies.
Question 72
An ML engineer has trained an ML model by using Amazon SageMaker AI. The ML engineer determines that the model is overfitting and that the training data contains unnecessary features. The ML engineer must reduce the overfitting and the impact of the unnecessary features.
Which solution will meet these requirements?
A. Increase the number of training iterations. Retrain the model.
B. Apply L1 regularization to the training data. Retrain the model.
C. Decrease the number of training iterations. Retrain the model.
D. Use SageMaker Debugger to apply L1 regularization to the running model.
Show Answer
Correct Answer: B
Explanation: L1 regularization (Lasso) penalizes the absolute magnitude of feature weights, encouraging irrelevant feature coefficients to become zero. This helps perform implicit feature selection and reduces overfitting. Increasing iterations can worsen overfitting, decreasing iterations is early stopping but does not specifically address unnecessary features, and SageMaker Debugger is for monitoring/debugging rather than applying regularization.
Question 73
An ML engineer at an insurance company trains a regression model to predict the number of insurance policy sales each month. After training the model, the ML engineer uses Amazon SageMaker AI to deploy the model for inference. The ML engineer wants to monitor the model predictions to detect whether the production data distribution differs from the training data distribution when there are changes in customer behaviors.
Which solution will meet these requirements?
A. Determine whether there is drift in the data quality.
B. Determine whether there is drift in the model quality.
C. Determine whether there is drift in the model bias.
D. Determine whether there is drift in the feature attribution.
Show Answer
Correct Answer: A
Explanation: The requirement is to detect whether the production data distribution differs from the training data distribution. In Amazon SageMaker Model Monitor, data quality monitoring detects data drift by comparing production input data against a baseline generated from the training data, including changes in feature distributions, missing values, and schema. Model quality monitors prediction performance against ground truth, bias monitors fairness metrics, and feature attribution monitors changes in feature importance rather than input distribution.
Question 74
An ML engineer uses one ML framework to train multiple ML models. The ML engineer needs to optimize the inference costs and host the models on Amazon SageMaker AI.
Which solution will meet these requirements MOST cost-effectively?
A. Create a multi-container inference endpoint for direct invocation.
B. Create a multi-model inference endpoint for all the models.
C. Create a multi-container inference endpoint for sequential invocation.
D. Create multiple single-model inference endpoint for each model.
Show Answer
Correct Answer: B
Explanation: A multi-model inference endpoint is the most cost-effective choice when hosting multiple models that use the same ML framework. It allows multiple models to share the same endpoint and compute resources, loading models on demand from Amazon S3, reducing infrastructure costs compared to separate endpoints. Multi-container endpoints are intended for hosting different containers or inference pipelines rather than multiple models of the same framework.
Question 75
A company runs a neural network model and retrains the model when the performance degrades. The company uses a training job that uses Amazon SageMaker AI distributed data parallelism (DDP). The training job takes several hours to run.
The company wants to decrease the required time for the training job.
Which solution will meet this requirement?
A. Increase the number of epochs.
B. Increase the number of neurons in the hidden layers.
C. Increase the number of layers.
D. Increase the number of instances.
Show Answer
Correct Answer: D
Explanation: Amazon SageMaker Distributed Data Parallelism speeds training by distributing batches across multiple instances and synchronizing gradients. If the job already uses DDP, increasing the number of instances increases parallelism and can reduce overall training time (subject to scaling efficiency). Increasing epochs, neurons, or layers would generally increase training time rather than decrease it.
Question 76
A travel company wants to create an ML model to recommend the next airport destination for its users. The company has collected millions of data records about user location, recent search history on the company’s website, and 2,000 available airports. The data has several categorical features with a target column that is expected to have a high-dimensional sparse matrix.
The company needs to use Amazon SageMaker AI built-in algorithms for the model. An ML engineer converts the categorical features by using one-hot encoding.
Which algorithm should the ML engineer implement to meet these requirements?
A. Use the CatBoost algorithm to recommend the next airport destination.
B. Use the DeepAR forecasting algorithm to recommend the next airport destination.
C. Use the Factorization Machines algorithm to recommend the next airport destination.
D. Use the k-means algorithm to cluster users into groups. Map each group to the next airport destination based on user search history.
Show Answer
Correct Answer: C
Explanation: Factorization Machines are designed for recommendation tasks with high-dimensional sparse feature spaces, such as one-hot encoded categorical variables. They efficiently model interactions between sparse features, making them well suited for predicting the next airport destination. CatBoost is a general gradient boosting algorithm, DeepAR is for time-series forecasting, and k-means is an unsupervised clustering algorithm rather than a recommendation model.
Question 77
A company uses an Amazon QuickSight dashboard to track the sale prices of sneakers over time. The dashboard aggregates sale prices scraped from many retail websites. The company wants to determine which prices are unusually high outliers and to display the outliers visually.
Which solution will meet these requirements?
A. Use a vertical bar chart to visualize the outliers. Use a calculated field in QuickSight to take the square roots of the outlier prices to generate the chart. Configure a custom AWS Lambda function to scan the data for anomalies.
B. Use AWS Glue DataBrew to preprocess the data. Set the REMOVE_OUTLIERS operation to eliminate data rows that include unusually high prices. Invoke an AWS Lambda function to store the removed rows in Amazon DynamoD
C. Use a vertical bar chart to visualize the outliers. Use a calculated field in QuickSight to square the outlier prices to generate the chart. Use QuickSight anomaly detection insights to determine which prices are unusually high.
D. Use a QuickSight filter to find the lowest 10 values for sneaker price. Assign a specific color to the 10 lowest values.
Show Answer
Correct Answer: C
Explanation: Amazon QuickSight supports ML-powered anomaly detection insights for time-series visualizations, which can identify unusually high or low values and highlight them visually. The other options either remove outliers instead of displaying them, rely on unnecessary custom Lambda processing, or focus on lowest values rather than unusually high outliers.
Question 78
A company uses an NFS-based data store to store data for ML training. Linux-based systems access the data store.
The company needs a hybrid system to make the shared data store accessible to on-premises servers and Amazon SageMaker AI notebooks that will consume the data. File locking is required for the data producers.
Which AWS storage solution will meet these requirements?
A. Use an Amazon S3 bucket to store the data. Use Mountpoint for Amazon S3 to mount the S3 bucket to the on-premises servers and the SageMaker AI notebooks.
B. Use an Amazon Elastic File System (Amazon EFS) file system to store the data. Mount the file system to the on-premises servers and the SageMaker AI notebooks.
C. Use an Amazon FSx for Lustre file system to store the data. Mount the file system to the on-premises servers and the SageMaker AI notebooks.
D. Use an Amazon Elastic Block Store (Amazon EBS) volume to store the data. Mount the volume to the on-premises servers and the SageMaker AI notebooks.
Show Answer
Correct Answer: B
Explanation: Amazon EFS is a managed NFS file system that supports Linux clients, POSIX semantics, and file locking. It can be mounted by SageMaker notebooks within a VPC and accessed from on-premises environments over AWS Direct Connect or VPN, making it suitable for a hybrid shared file system. S3 with Mountpoint does not provide full POSIX semantics or file locking, FSx for Lustre is not the standard hybrid NFS solution for this use case, and EBS cannot be shared across on-premises systems and SageMaker as a network file system.
Question 79
An ML engineer is building an ML pipeline. The pipeline must process a dataset in two ways by using Amazon Athena. The pipeline must use batch processing to perform large-scale data transformations and for model training. The pipeline must also use near real-time processing to perform low-latency queries for inference and analytics.
Which file format will provide the LEAST latency for both types of processing?
A. CSV
B. Apache Parquet
C. Nested JSON
D. Deserialized JSON
Show Answer
Correct Answer: B
Explanation: Apache Parquet is a columnar storage format optimized for analytics. It provides efficient compression, predicate pushdown, and reads only the required columns, making Athena queries much faster for both large-scale batch transformations and low-latency analytical queries compared with CSV or JSON formats.
Question 80
A company is training a new ML model to replace a model that is deployed on an Amazon SageMaker AI real-time endpoint. An ML engineer needs to determine the latency and the accuracy of the new model. The ML engineer must evaluate the new model in a production scenario without affecting the users of the existing model.
Which solution will meet these requirements?
A. Perform a blue/green deployment with linear traffic shifting.
B. Perform a blue/green deployment with canary traffic shifting.
C. Perform a rolling deployment with a rolling batch size of 50% of the current fleet.
D. Perform shadow testing with a traffic sampling percentage of 100%.
Show Answer
Correct Answer: D
Explanation: Shadow testing duplicates production requests to the new model while continuing to serve responses from the existing model. This allows measurement of latency and accuracy under real production traffic without impacting users. Blue/green and rolling deployments shift live user traffic to the new model, which can affect users.
$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.
