Free Amazon MLA-C01 Practice Exams - Page 3 of 6

Question #21

Which AWS services are specifically designed to aid in monitoring machine learning models and incorporating human review processes? (Select two)

A . Amazon SageMaker Feature Store
B . Amazon SageMaker Ground Truth
C . Amazon Augmented AI (Amazon A2I)
D . Amazon SageMaker Data Wrangler
E . Amazon SageMaker Model Monitor

Reveal Solution Hide Solution

Question #22

You are preparing a dataset for training a machine learning model using SageMaker Data Wrangler. The dataset has several missing values spread across different columns, and these columns contain numeric data. Before training the model, it is essential to handle these missing values to ensure the model performs optimally. The goal is to replace the missing values in each numeric column with the mean of that column.

Which transformation in SageMaker Data Wrangler should you apply to replace the missing values in numeric columns with the mean of those columns?

A . Encode
B . Impute
C . Scale
D . Drop

Reveal Solution Hide Solution

Question #23

Why is it important to estimate the amount of uncertainty in an ML model?

A . To improve the model’s computational efficiency
B . To increase the speed of data processing
C . To reduce the amount of training data needed
D . To avoid potential misinterpretations of life and property

Reveal Solution Hide Solution

Question #24

What characterizes a simple machine learning model?

A . Low accuracy but high explainability
B . High accuracy and high explainability
C . Low accuracy and low explainability
D . High accuracy but low explainability

Reveal Solution Hide Solution

Question #25

You are a machine learning engineer at a healthcare startup that uses an Amazon SageMaker endpoint to deliver real-time diagnostics based on patient data. The model needs to handle a high volume of requests with low latency to ensure timely results. Recently, the startup has experienced rapid growth, leading to occasional periods of high traffic where users experience increased latency and, in some cases, request timeouts. You also need to be mindful of cost, as the startup operates on a tight budget.

Which approach is the MOST EFFECTIVE for troubleshooting and resolving the capacity concerns while balancing cost and performance?

A . Request a service quota increase for the SageMaker endpoint to allow for more instances during peak traffic, and set up a CloudWatch Alarm to notify you when utilization exceeds 80% of the current quota
B . Increase the instance size for the SageMaker endpoint to handle more requests per instance, and manually monitor performance and costs using Amazon CloudWatch metrics
C . nable auto-scaling for the SageMaker endpoint to automatically adjust the number of instances based on request load, and set a budget alert in AWS Budgets to monitor cost increases as traffic scales
D . Use AWS Lambda with provisioned concurrency to handle the requests, ensuring that the function is always ready to serve traffic. Configure Lambda to auto-scale based on traffic but limit the maximum concurrency to control costs

Reveal Solution Hide Solution

Correct Answer: C
C

Explanation:

Correct option:

nable auto-scaling for the SageMaker endpoint to automatically adjust the number of instances based on request load, and set a budget alert in AWS Budgets to monitor cost increases as traffic scales

Enabling auto-scaling on the SageMaker endpoint allows the system to automatically adjust the number of instances based on incoming traffic, ensuring that it can handle spikes without degrading performance. With a target tracking scaling policy, you choose an Amazon CloudWatch metric and target value. Auto scaling creates and manages the CloudWatch alarms for the scaling policy and calculates the scaling adjustment based on the metric and the target value. The policy adds and removes the number of instances as required to keep the metric at, or close to, the specified target value. For example, a scaling policy that uses the predefined InvocationsPerInstance metric with a target value of 70 can keep InvocationsPerInstance at, or close to 70.

By setting up a budget alert in AWS Budgets, you can monitor cost increases and ensure that scaling does not exceed your budget. This approach provides a balance between performance and cost, allowing the system to scale dynamically while keeping expenses in check.

via – https://docs.aws.amazon.com/cost-management/latest/userguide/budgets-managing-costs.html

Incorrect options:

Increase the instance size for the SageMaker endpoint to handle more requests per instance, and manually monitor performance and costs using Amazon CloudWatch metrics – Increasing the instance size might temporarily address performance issues, but it is less flexible and could lead to over-provisioning, resulting in unnecessary costs during periods of low traffic. Manual monitoring is also less efficient and can lead to delays in addressing issues.

Use AWS Lambda with provisioned concurrency to handle the requests, ensuring that the function is always ready to serve traffic. Configure Lambda to auto-scale based on traffic but limit the maximum concurrency to control costs – Provisioned concurrency is a feature that keeps Lambda functions initialized and hyper-ready to respond in double-digit milliseconds. This is ideal for implementing interactive services, such as web and mobile backends, latency-sensitive microservices, or synchronous APIs. Using AWS Lambda with provisioned concurrency is suitable for serverless workloads, but it may not be the best fit for a SageMaker model that requires consistent performance at scale. Additionally, limiting concurrency to control costs could lead to performance bottlenecks during high-traffic periods. Request a service quota increase for the SageMaker endpoint to allow for more instances during peak traffic, and set up a CloudWatch Alarm to notify you when utilization exceeds 80% of the current quota –

Your AWS account has default quotas, formerly referred to as limits, for each AWS service. Requesting a service quota increase can help if the current quota is a limiting factor, but it doesn’t directly address cost control or the need for dynamic scaling. It’s a reactive approach that doesn’t provide the flexibility of auto-scaling combined with budget monitoring.

References:

https://docs.aws.amazon.com/sagemaker/latest/dg/endpoint-auto-scaling-prerequisites.html

https://docs.aws.amazon.com/cost-management/latest/userguide/budgets-managing-costs.html

https://aws.amazon.com/blogs/aws/new-provisioned-concurrency-for-lambda-functions/

https://docs.aws.amazon.com/general/latest/gr/aws_service_limits.html

Question #26

You are an ML Engineer working for a healthcare company that uses a machine learning model to recommend personalized treatment plans to patients. The model is deployed on Amazon SageMaker and is critical to the company’s operations, as any incorrect predictions could have significant consequences. A new version of the model has been developed, and you need to deploy it in production. However, you want to ensure that the deployment process is robust, allowing you to quickly roll back to the previous version if any issues arise. Additionally, you need to maintain version control for future updates and manage traffic between different model versions.

Which of the following strategies should you implement to ensure a smooth and reliable deployment of the new model version using Amazon SageMaker, considering best practices for versioning and rollback strategies? (Select two)

A . Utilize Amazon SageMaker’s blue/green deployment strategy to shift traffic gradually from the old model to the new one, ensuring that you can monitor performance and quickly revert if needed
B . Deploy the new model version immediately and redirect 100% of traffic to it, assuming it has been thoroughly tested and will not require a rollback
C . Use Amazon SageMaker’s built-in versioning to manage different versions of the model, and deploy the new version in a canary release by redirecting a small percentage of traffic to it initially
D . Create a backup of the current model, deploy the new version, and if any issues arise, manually roll back by redeploying the previous model version
E . Deploy the new model version alongside the current one, and use Amazon SageMaker’s multi-
model endpoint to serve both models simultaneously, splitting traffic between them

Reveal Solution Hide Solution

Correct Answer: A, C
A, C

Explanation:

Correct options:

Use Amazon SageMaker’s built-in versioning to manage different versions of the model, and deploy the new version in a canary release by redirecting a small percentage of traffic to it initially Amazon SageMaker supports model versioning, which is crucial for tracking different iterations of your model. A canary release allows you to deploy the new model version to a small portion of users, minimizing risk by limiting exposure in case of issues. If the new version performs well, you can gradually increase traffic to it.

Utilize Amazon SageMaker’s blue/green deployment strategy to shift traffic gradually from the old

model to the new one, ensuring that you can monitor performance and quickly revert if needed

A blue/green deployment strategy is a best practice in model deployment. It allows you to deploy the new

model version in parallel with the existing one, gradually shifting traffic to the new version while monitoring

its performance. If issues are detected, you can quickly roll back to the previous version without disrupting

service.

In a blue/green deployment, SageMaker provisions a new fleet with the updates (the green fleet). Then, SageMaker shifts traffic from the old fleet (the blue fleet) to the green fleet. Once the green fleet operates smoothly for a set evaluation period (called the baking period), SageMaker terminates the blue fleet. You can specify Amazon CloudWatch alarms that SageMaker uses to monitor the green fleet. If an issue with the updated code trips any of the alarms, SageMaker initiates an auto-rollback to the blue fleet in order to maintain availability thereby minimizing risk.

via – https://docs.aws.amazon.com/sagemaker/latest/dg/deployment-guardrails-blue-green.html

Incorrect options:

Deploy the new model version immediately and redirect 100% of traffic to it, assuming it has been thoroughly tested and will not require a rollback – Redirecting 100% of traffic to the new model version immediately is risky, especially in a critical application like healthcare. Without a rollback plan, any issues with the new version could lead to significant consequences.

Create a backup of the current model, deploy the new version, and if any issues arise, manually roll back by redeploying the previous model version – While creating a backup and manually rolling back is a possible strategy, it is not as efficient or reliable as using a built-in rollback feature like blue/green deployment or canary releases. Manual rollbacks can lead to delays and increased downtime.

Deploy the new model version alongside the current one, and use Amazon SageMaker’s multi-model endpoint to serve both models simultaneously, splitting traffic between them – While using a

multi-model endpoint could serve both models simultaneously, it is not the best approach for managing risk during deployment. This strategy is more suited for scenarios where you need to serve multiple models for different purposes rather than managing a controlled rollout.

References:

https://docs.aws.amazon.com/sagemaker/latest/dg/deployment-guardrails-blue-green.html

https://docs.aws.amazon.com/sagemaker/latest/dg/deployment-guardrails-rolling.html

https://docs.aws.amazon.com/sagemaker/latest/dg/multi-model-endpoints.html

Question #27

What is the main goal of testing the model in ML?

A . To prepare the data for training
B . To generate predictions for new data
C . To construct input representations from raw variables
D . To evaluate the quality of the model on held-out data

Reveal Solution Hide Solution

Question #28

A company is using a fleet of Amazon EC2 instances to ingest data from on-premises data sources. The data is in JSON format and ingestion rates can be as high as 1 MB/s. When an EC2 instance is rebooted, the data in flight is lost. The company’s data science team wants to query ingested data in near-real time.

Which solution provides near-real-time data querying that is scalable with minimal data loss?

A . Store ingested data in an Amazon Elastic Block Store (Amazon EBS) volume. Write to Amazon S3 bucket and use Amazon Redshift Spectrum to query the data
B . Store ingested data in an Amazon Elastic Block Store (Amazon EBS) volume. Publish data to Amazon ElastiCache for Redis. Subscribe to the Redis channel to query the data
C . Publish data to Amazon Kinesis Data Streams, use Kinesis Data Analytics to query the data
D . Store ingested data in an EC2 instance store. Publish data to Amazon Kinesis Data Firehose with
Amazon S3 as the destination. Use Amazon Athena to query the data

Reveal Solution Hide Solution

Correct Answer: C
C

Explanation:

Correct option:

Publish data to Amazon Kinesis Data Streams, use Kinesis Data Analytics to query the data

With Amazon Kinesis Data Analytics for SQL Applications, you can process and analyze streaming data using standard SQL. The service enables you to quickly author and run powerful SQL code against streaming sources to perform time series analytics, feed real-time dashboards, and create real-time metrics.

To get started with Kinesis Data Analytics, you create a Kinesis Data Analytics application that continuously reads and processes streaming data. The service supports ingesting data from Amazon Kinesis Data Streams and Amazon Data Firehose streaming sources. Kinesis Data Analytics supports Amazon Data Firehose (Amazon S3, Amazon Redshift, Amazon OpenSearch Service, and Splunk), AWS Lambda, and Amazon Kinesis Data Streams as destinations. Kinesis Data Stream and Kinesis Data Analytics:

via – https://aws.amazon.com/kinesis/

Incorrect options:

Store ingested data in an Amazon Elastic Block Store (Amazon EBS) volume. Publish data to Amazon ElastiCache for Redis. Subscribe to the Redis channel to query the data

Store ingested data in an Amazon Elastic Block Store (Amazon EBS) volume. Write to Amazon S3 bucket and use Amazon Redshift Spectrum to query the data

Amazon Elastic Block Store (EBS) is not typically used for streaming data. EBS is designed as block-level storage for use with Amazon EC2 instances, providing persistent storage that remains available independent of the lifecycle of the EC2 instance. It is ideal for workloads requiring consistent and low-latency storage performance, such as databases, file systems, and applications that need block storage. Therefore, both these options are incorrect.

Store ingested data in an EC2 instance store. Publish data to Amazon Kinesis Data Firehose with Amazon S3 as the destination. Use Amazon Athena to query the data – The instance store is temporary

storage and the data will be lost when the EC2 instance reboots.

References:

https://docs.aws.amazon.com/kinesisanalytics/latest/dev/what-is.html https://repost.aws/questions/QUpj1ciWvKTfCN8UGJ-UdqnQ/kinesis-data-ingestion

Question #29

How does the Amazon flywheel model ensure continuous improvement and growth?

A . By using feedback loops that integrate customer experience, selection, and pricing.
B . By limiting the scope of operational improvements to cost reduction only.
C . By avoiding changes to business practices.
D . By maintaining a static approach to business operations.

Reveal Solution Hide Solution

Question #30

You are a machine learning engineer at a biotechnology company working on a deep learning model to analyze genomic sequences. The model requires significant computational resources for training due to the complexity and size of the dataset, which consists of billions of nucleotide sequences. Additionally, once deployed, the model must provide real-time inferences for clinical applications, requiring low-latency predictions. Your task is to choose the appropriate compute environment for both training and inference.

Which of the following compute environment configurations is the MOST SUITABLE for meeting the training and inference requirements of this use case?

A . Use Amazon SageMaker with p3 instances (NVIDIA V100 GPUs) for training, and deploy the model using ml.m5.large instances (Intel Skylake CPUs) for real-time inference to balance cost and performance
B . Use Amazon EC2 with g4dn instances (NVIDIA T4 GPUs) for both training and inference, optimizing for cost-efficiency while maintaining moderate GPU performance
C . Use Amazon SageMaker with ml.c5.large instances (Intel Cascade Lake CPUs) for training and inference, relying on the CPU’s high memory bandwidth for both tasks
D . Use Amazon SageMaker with p4d instances (NVIDIA A100 GPUs) for training to handle large-scale deep learning workloads, and deploy the model using ml.inf1 instances (AWS Inferentia chips) for low-latency inference

Reveal Solution Hide Solution

Correct Answer: D
D

Explanation:

Correct option:

Use Amazon SageMaker with p4d instances (NVIDIA A100 GPUs) for training to handle large-scale deep learning workloads, and deploy the model using ml.inf1 instances (AWS Inferentia chips) for low-latency inference

p4d instances with NVIDIA A100 GPUs are optimized for large-scale deep learning workloads, providing the necessary computational power for training complex models on large datasets. For inference, ml.inf1 instances with AWS Inferentia chips are designed specifically for high-performance, low-latency inference, making this combination ideal for both training and real-time clinical applications.

Incorrect options:

Use Amazon SageMaker with p3 instances (NVIDIA V100 GPUs) for training, and deploy the model using ml.m5.large instances (Intel Skylake CPUs) for real-time inference to balance cost and performance – While p3 instances with NVIDIA V100 GPUs are powerful for deep learning training, ml.m5.large instances with Intel Skylake CPUs may not deliver the low-latency inference needed for real-time clinical applications. The CPU-based inference environment could be a bottleneck.

Use Amazon SageMaker with ml.c5.large instances (Intel Cascade Lake CPUs) for training and inference, relying on the CPU’s high memory bandwidth for both tasks – ml.c5.large instances with Intel Cascade Lake CPUs are suitable for general-purpose compute tasks, but they lack the specialized GPU performance needed for deep learning training and the low-latency capabilities required for real-time inference.

Use Amazon EC2 with g4dn instances (NVIDIA T4 GPUs) for both training and inference, optimizing for cost-efficiency while maintaining moderate GPU performance – g4dn instances with NVIDIA T4 GPUs are cost-efficient and suitable for moderate workloads, but they may not provide the high-performance training environment needed for a large genomic dataset. Additionally, they might not meet the low-latency inference requirements as effectively as ml.inf1 instances.

References:

https://aws.amazon.com/ec2/instance-types/

https://docs.aws.amazon.com/sagemaker/latest/dg/notebooks-available-instance-types.html