Machine Learning

Question 1

A computer scientist working for a university is looking to build a machine learning application which will use telemetry data to predict weather for a given area at a given time. This application would benefit from using managed services and will need to find a solution which uses third party data within the application.

Which of the following combinations of services will deliver the best solution?

1. Use Amazon SageMaker to build the machine learning part of the application and use AWS DataSync to gain access to the third-party telemetry data.
2. Use a TensorFlow AMI from the AWS Marketplace to build the machine learning part of the application and use AWS DataSync to gain access to the third-party telemetry data.
3. Use a TensorFlow AMI from the AWS Marketplace to build the machine learning part of the application and use AWS Data Exchange to gain access to the third-party telemetry data.
4. Use Amazon SageMaker to build the machine learning part of the application and use AWS Data Exchange to gain access to the third-party telemetry data.

Answer 1

4. Use Amazon SageMaker to build the machine learning part of the application and use AWS Data Exchange to gain access to the third-party telemetry data.

Amazon SageMaker allows you to build, train, and deploy machine learning models for any use case with fully managed infrastructure, tools, and workflows. AWS Data Exchange allows you to gain access to third party data sets across Automotive, Financial Services, Gaming, Healthcare & Life Sciences, Manufacturing, Marketing, Media & Entertainment, Retail, and many more industries.

• AWS DataSync is a secure, online service that automates and accelerates moving data between on-premises and AWS storage services. It does not give access to third party data.
• Building an EC2 instance from a TensorFlow AMI would not involve using managed services and AWS DataSync is a secure, online service that automates and accelerates moving data between on-premises and AWS storage services. It does not give access to third party data.
• Building an EC2 instance from a TensorFlow AMI would not involve using managed services.

Reference:
AWS Data Exchange

Question 2

A financial institution wants to use machine learning (ML) algorithms to detect potential fraudulent transactions. They need to create ML models based on their vast financial transaction data and integrate these models into their business intelligence system for real-time decision-making. The solution should require minimal operational overhead.

Which solution will best meet these requirements?

1. Use Amazon SageMaker to build, train, and deploy ML models, and use Amazon QuickSight for data visualization.
2. Use AWS Glue to perform ETL jobs on the transaction data and use Amazon Forecast for predictive analytics.
3. Use a pre-built ML Amazon Machine Image (AMI) from the AWS Marketplace to build and train models and use AWS Athena for data visualization.
4. Use Amazon Comprehend for analyzing the transaction data and Amazon Elasticsearch for visualization.

Answer 2

1. Use Amazon SageMaker to build, train, and deploy ML models, and use Amazon QuickSight for data visualization.

Amazon SageMaker is a fully managed service that provides every developer and data scientist with the ability to build, train, and deploy machine learning models quickly. It can directly connect with data sources and has built-in algorithms to ease the ML process.

Amazon QuickSight is a business intelligence tool that can be used to create dashboards for data visualization. This combination perfectly suits the requirement.

• AWS Glue is primarily used for ETL jobs - cleaning, preparing, and moving data. Amazon Forecast is a fully managed service for time-series forecasting, which might not be a complete solution for detecting fraudulent transactions.
• AWS Marketplace ML AMIs can be used to create and train models, but this will require manual operational effort in terms of setting up and managing the instances. Athena is a query service and does not provide data visualization capabilities that a business intelligence tool like QuickSight provides.
• Amazon Comprehend is primarily used for natural language processing (NLP), which isn’t suited for detecting fraudulent transactions. Elasticsearch is a search and analytics engine and might not be the best tool for the use case described here.

Reference:
Amazon SageMaker

Save time with our AWS cheat sheets.