AWS MACHINE LEARNING QUIZ SET 1 Author: CloudVikas Published Date: 28 April 2020 Welcome to AWS MACHINE LEARNING QUIZ SET 1. Please enter your email details to get QUIZ Details on your email id. Click on Next Button to proceed. Email 1. ######QUESTION#########Which best describes A/B Testing? (Choose 2) For a period of time, we're going to be running two customer databases and manually syncing records between them until we can upgrade the other portions of our eCommerce site. We will be rolling out our new customer recommendation engine to a small percentage of customers to evaluate the results. Because our order processing system is loosely coupled with our inventory system, we will be temporarily queueing orders over the weekend as we upgrade the inventory system. Then, next weekend, we will upgrade the order processing system. Because our inventory and accounting systems are very tightly coupled, we will be upgrading both systems over the holiday weekend.2. ######QUESTION#########Which one of these is not a common machine learning programming language?Jupyter NotebookJuliaR3. ######QUESTION#########As a ML engineer , John got requirement to deliver ML framework.He has got few ML frameworks option to choose.Which of these is a very popular machine learning framework?ScalaPyTorchPython4. ######QUESTION#########What is a major economic benefit of cloud computing for businesses?Cloud computing resources depreciate at a faster rate than in-house resources.Cloud computing platforms are simpler to use than in-house resources.Computing resources can be procured on an OpEx basis versus a CapEx basis.5. ######QUESTION#########Which statement is accurate about architecting for the cloud versus in-house landscapes?Hybrid architectures are only advisable when cloud computing costs are more than in-house costs.Architecting for the cloud is a discipline itself with its own best practices and complexities.Architecting a cloud solution is no different than architecting an in-house solution.6. ######QUESTION#########Which of the following is NOT a reason cloud computing is often used for machine learning today?The distributed nature of cloud computing allows for less expensive model development.Cloud providers have made available easy access to many machine learning services without the need to create whole models from scratch.Cloud computing allows for quick access to sophisticated equipment.7. ######QUESTION#########Which best describes Concept Drift?The performance of your machine learning model has increased as computing power gets cheaper.Your machine learning model has gotten better over time.The profile of the underlying ground truth data used in your model training had changed over time.8. ######QUESTION#########Which of the following statements about computing hardware and ML are not true?ASICS are not available to the general public.GPUs are specialized at performing math functions very fast.CPUs and GPUs are the most flexible device for machine learning development and iteration.9. ######QUESTION#########Why is the Trolley Dilemma a favorite discussion topic with it comes to machine learning?It helps consider what sort of responsibilities we might have to allow algorithms and models to make on our behalf.It aids in discussions with other project members around various bias that could get introduced into our models.It helps frame the challenge of risk management in machine learning projects.10. ######QUESTION#########Which best describes a Big Bang deployment strategy?For a period of time, we're going to be running two customer databases and manually syncing records between them until we can upgrade the other portions of our eCommerce site.We will be rolling out our new customer recommendation engine to a small percentage of customers to evaluate the results.Because our inventory and accounting systems are very tightly coupled, we will be upgrading both systems over the holiday weekend.11. ######QUESTION#########As you know,Data bias in machine learning is a type of error in which certain elements of a dataset are more heavily weighted and/or represented than others.What stage in the machine learning project process should you consider potential bias?Problem Framing stageTrainingData Preparation12. ######QUESTION#########You conduct a survey via your website to ask what type of blog articles people most want. Your normal traffic is 100 visitors a day but you receive 4 responses. What type of bias might be present here?Sampling BiasSymbolic BiasParticipation Bias13. ######QUESTION#########If I were to assume that all dogs bark because that's what I've always been told, what type of bias might this be?Implicit BiasExperimenter's BiasConvenience Bias14. ######QUESTION#########Which phase in an ML project will likely consume the most time and resources?Data PreparationModel ValidationModel Deployment15. ######QUESTION#########What deployment strategy might you recommend if you were unsure how your model might perform under the heavy load in production?Big Bang DeploymentRapid Application DeploymentCanary Deployment16. ######QUESTION#########If you wanted to create a device that was purpose built for one specific machine learning task and didn't care about upgrading it later, which type of device might you choose if raw speed was your only concern?ASICGPUFPGA17. ######QUESTION#########Which of the following will likely be most effective in incorporating the topic of ethics in your ML project?Conduct a Worst Case Workshop.Insist that the local newspaper runs a story on your new ML model.Implementing an Ethics Policy and requiring project members to sign it.18. ######QUESTION#########We decide to use a dataset that is 10 years old to evaluate student preferences for online coaching. A newer dataset is available, but it would require us to download 150 different files rather than the 1 file of the old data.. What bias might be be introducing?Correlation BiasConvenience BiasAutomation Bias19. ######QUESTION#########Which of the following might be considered under the risk heading of Interpretation?Your organization intends to disband the ML project team as soon as the model is in production.You are upset at the weather website for saying it was going to be sunny but instead it rained.Your model consistently returns errors when the input data happens to be very high numbers.20. ######QUESTION#########Which would you recommend to someone just starting their machine learning journey? Start small and simpleDon't start. The less people know about ML, the better. It will just scare people.Don't embarrass yourself by asking questions until you have at least 2 years of experience.21. ######QUESTION#########Which is an example of online validation? (Choose 2) Creating a copy of the production database and running the model through the historical data. Deploying an A/B test for your new model. Testing the model using the QA version of the database rather than the production version. Running the model as a shadow deployment for a couple weeks.22. ######QUESTION#########In reviewing the testing results from your new ML model, you notice that results are very consistent but that they are consistently 20% higher than they should be. What might you say about your model?Low Accuracy, High PrecisionHigh Accuracy, Low PrecisionHigh Accuracy, High Precision23. ######QUESTION#########Which of the following would most impact how you framed an ML problem? You will need to scale some values in the data using a normalization method.You discover the data you intend to use was stolen from a competitor.You realize the data is in metric units but your model intends to produce output in imperial units.24. ######QUESTION#########Which best describes a requirement in our ML project charter to include some time to ensure we can show an external auditor how our model makes decisions?IntegrityConsistencyExplainability25. ######QUESTION#########A project member insists that we keep training our model until it produces 100% perfect results in testing. What risk might this introduce?OverfittingImplicit BiasOver-planning26. ######QUESTION#########If we assume that all Canadians also speak French, what type of bias might this be?Group Attribution BiasSampling BiasAutomation Bias27. ######QUESTION#########As a ML engineer , you have to develop ML framework. An ML framework is any tool, interface, or library that lets you develop ML models easily. Which are benefits of using a framework for your machine learning project? Can help you do things faster. Pre-made collection of resources. Provides a way to document your code. Ensures your models are highly accurate.28 out of 27Please fill in the comment box below. Author: CloudVikas
