Introduction to Machine Learning in Production Coursera Quiz Answers - Courseinside (2023)

Here you will get Introduction to Machine Learning in Production Coursera Quiz Answers

In the first course of Machine Learning Engineering for Production Specialization, you will identify the various components and design an ML production system end-to-end: project scoping, data needs, modeling strategies, and deployment constraints and requirements; and learn how to establish a model baseline, address concept drift, and prototype the process for developing, deploying, and continuously improving a productionized ML application

SKILLS YOU WILL GAIN

• Human-level Performance (HLP)
• Concept Drift
• Model baseline
• Project Scoping and Design
• ML Deployment Challenges

All Quiz Answers For Week 1

Quiz 1 >>The Machine Learning Project Lifecycle

Q1. Which of these are stages of the machine learning project lifecycle? Check all that apply.

1. Data
2. Configuration
3. Modeling
4. Scoping
5. Deployment

Q2. Which of these is not an advantage of a typical edge deployment compared to a typical cloud deployment?

1. Less network bandwidth needed
2. Lower latency
3. Can function even if network connection is down
4. More computational power available

Q3. In the speech recognition example, what is the problem with some labelers transcribing audio as “Um, today’s weather” and others transcribing “Umm…, today’s weather”?

1. We should not be transcribing “Umm.” The correct transcription, which serves the user’s needs better, is just “Today’s weather.
2. The first is grammatically incorrect and we should use the second transcription.
3. Either transcription is okay, but the inconsistency is problematic.
4. The second is grammatically incorrect and we should use the first transcription.

Q4. After a system is deployed, monitoring and maintaining the system will help us handle cases of concept drift or data drift.

1. True
2. False

Q5. Which statement is a more accurate description of the full cycle of a machine learning project?

1. It is a linear process, in which we move step-by-step from scoping to deployment. (That’s why we call it a cycle. Bicycles are only good at going forward, not backward.)
2. It is an iterative process, where during a later stage we might go back to an earlier stage. (That’s why we call it a cycle–it’s a circular process.)

Quiz 2 >>Deployment

Q1. You’ve built a new system for making loan approval decisions. For now, its output is not used in any decision making process, and a human loan officer is solely responsible for deciding what loans to approve. But the system’s output is logged for analysis. What is this type of deployment called?

1. Shadow mode deployment
2. Red green deployment
3. Canary deployment
4. Blue green deployment

Q2. On a new social media platform, you’re rolling out a new anti-spam system to flag and hide spammy posts. Your team decides to roll out the anti-spam filter via a canary deployment, and roll it out to 1% of users initially. Which of these would you advocate?

1. Monitor that 1% of users’ reaction, and if it goes well, flip the switch to send all traffic (100%) to the system.
2. Use a plan to ramp up to more users at a fixed rate: 1% in the first week, 2% in second week, 4% in third, and so on, so that the rollout can be well planned and managed.
3. Monitor that 1% of users’ reaction, and either gradually ramp up (if it’s going well) or rollback (if not)
4. After a successful canary deployment, begin to implement a shadow mode deployment.

Q3. You’re building a healthcare screening system, where you input a patient’s symptoms, and for the easy cases (such as an obvious case of the common cold) the system will give a recommendation directly, and for the harder cases it will pass the case on to a team of in-house doctors who will form their own diagnosis independently. What degree of automation are you implementing in this example for patient care?

1 point

1. Human only
2. Full Automation
3. Partial Automation
4. Shadow mode

Q4. You have built and deployed an anti-spam system that inputs an email and outputs either 0 or 1 based on whether the email is spam. Which of these will result in either concept drift or data drift?

1 point

1. Spammers trying to change the wording used in emails to get around your spam filter.
2. Cloud computational costs going down, resulting in a lower cost to process each email received.
3. Updating a monitoring dashboard to keep track of new metrics.
4. None of these will result in either concept drift or data drift.

Q5. Which of these statements is a more accurate description of deployment?

1. It is an iterative process, where you should expect to make multiple adjustments (such as metrics monitored using dashboards or percentage of traffic served) to work towards optimizing the system.
2. Because deployment is a high stakes event, it’s critical to design the right system, so that immediately after launch it will immediately work reliably and scale effectively.

All Quiz Answers For Week 2

Quiz 1 >>Selecting and Training a Model

Q1. Which of these is a more accurate description of a data-centric approach to ML development?

1. Holding the training data fixed, work to improve your neural network’s architecture to do well on the problem.
2. Holding the neural network architecture fixed, work to improve the data to do well on the problem.

Q2. Say you have an algorithm that diagnoses illnesses from medical X-rays, and achieves high average test set accuracy. What can you now say with high confidence about this algorithm? Check all that apply.

1. It does well even on rare classes of diseases.
2. Its diagnoses are roughly equally accurate on all genders and ethnicities, so we are confident it is not biased against any gender or ethnicity.
3. The system can be safely deployed in a healthcare setting.
4. None of the above.

Q3. Which of these statements about establishing a baseline are accurate? Check all that apply.

1. Human level performance (HLP) is generally more effective for establishing a baseline on unstructured data problems (such as images and audio) than structured data problems
2. Open-source software should not be used to establish a baseline, since the performance of a good open source implementation might be too good and thus too hard to beat.
3. For unstructured data problems, using human-level performance as the baseline can give an estimate of the irreducible error/Bayes error and what performance is reasonable to achieve.
4. It can be established based on an older ML system

Q4. On a speech recognition problem, say you run the sanity-check test of trying to overfit a single training example. You pick a clearly articulated clip of someone saying “Today’s weather”, and the algorithm fails to fit even this single audio clip, and outputs “______”. What should you do?

1. Train the algorithm on a larger dataset to help it to fit the data better.
2. Use data augmentation on this one audio clip to make sure the algorithm hears a variety of examples of “today’s weather” to fit this phrase better.
3. Debug the code/algorithm/hyperparameters to make it pass this sanity-check test first, before moving to larger datasets.
4. Create a training set of this example repeated 100 times to force the algorithm to learn to fit this example well.

Quiz 2 >>Modeling challenges

Q1. You are working on a binary classification ML algorithm that detects whether a patient has a specific disease. In your dataset, 98% of the training examples (patients) don’t have the disease, so the dataset is very skewed. Accuracy on both positive and negative classes is important. You read a research paper claiming to have developed a system that achieves 95% on ____ metric. What metric would give you the most confidence they’ve built a useful and non-trivial system? (Select one)

1 point

1. Recall
2. Accuracy
3. F1 score
4. Precision

Q2. On the previous problem above with 98% negative examples, if your algorithm is print(“1”) (i.e., it says everyone has the disease). Which of these statements is true?

1. The algorithm achieves 100% recall.
2. The algorithm achieves 100% precision.
3. The algorithm achieves 0% precision.
4. The algorithm achieves 0% recall.

Q3. True or False? During error analysis, each example should only be assigned one tag. For example, in a speech recognition application you may have the tags: “car noise”, “people noise” and “low bandwidth”. If you encounter an example with both car noise and low bandwidth audio, you should use your judgement to assign just one of these two tags rather than apply both tags.

1 point

1. True
2. False

Q4. You are building a visual inspection system. Error analysis finds:

Based on this, what is the more promising type of defect to work on?

1 point

1. Discoloration, because the algorithm’s accuracy is lower and thus there’s more room for improvement.
2. Discoloration, because HLP is lower which suggests this is therefore the harder problem that thus needs more attention.
3. Scratch defects, because the gap to HLP is higher and thus there’s more room for improvement.
4. Work on both classes equally because they are each 50% of the data.

Q5. You’re considering applying data augmentation to a phone visual inspection problem. Which of the following statements are true about data augmentation? (Select all that apply)

1 point

1. Data augmentation should try to generate more examples in the parts of the input space where the algorithm is already doing well and there’s no need for improvement.
2. Data augmentation should distort the input sufficiently to make sure they are hard to classify by humans as well.
3. GANs can be used for data augmentation.
4. Data augmentation should try to generate more examples in the parts of the input space where you’d like to see improvement in the algorithm’s performance.

All Quiz Answers For week 3

Quiz >>Data Stage of the ML Production Lifecycl

Q1. Which of these statements do you agree with regarding structured vs. unstructured data problems?

1. It is generally easier for humans to label data and to apply data augmentation on unstructured data than structured data.
2. It is generally easier for humans to label data and to apply data augmentation on structured data than unstructured data.
3. It is generally easier for humans to label data on unstructured data, and easier to apply data augmentation on structured data.
4. It is generally easier for humans to label data on structured data, and easier to apply data augmentation on unstructured data.

Q2. Take speech recognition. Some labelers transcribe with “…” (as in, “Um… today’s weather”) whereas others do so with commas “,”. Human-level performance (HLP) is measured according to how well one transcriber agrees with another. You work with the team and get everyone to consistently use commas “,”. What effect will this have on HLP?

1. HLP will decrease.
2. HLP will increase.
3. HLP will stay the same.

Q3. Take a phone visual inspection problem. Suppose even a human inspector looking at an image cannot tell if there is a scratch. If however the same inspector were to look at the phone directly (rather than an image of the phone) then they can clearly tell if there is a scratch. Your goal is to build a system that gives accurate inspection decisions for the factory (not publish a paper). What would you do?

1 point

1. Try to improve the consistency of the labels, y.
2. Get a big dataset of many training examples, since this is a challenging problem that will require a big dataset to do well on.
3. Carefully measure HLP on this problem (which will be low) to make sure the algorithm can match HLP.
4. Try to improve their imaging (camera/lighting) system to improve the quality or clarity of the input images, x.

Q4. You are building a system to detect cats. You ask labelers to please “use bounding boxes to indicate the position of cats.” Different labelers label as follows:

What is the most likely cause of this?

1. Labelers have not had enough coffee.
2. Ambiguous labeling instructions.
3. Lazy labelers.
4. That this should have been posed as a segmentation rather than a detection task.

Q5. You are building a visual inspection system. HLP is measured according to how well one inspector agrees with another. Error analysis finds:

You decide that it might be worth checking for label consistency on both scratch and discoloration defects. If you had to pick one to start with, which would you pick?

1. It is more promising to check (and potentially improve) label consistency on scratch defects than discoloration defects, since HLP is higher on scratch defects and thus it’s more reasonable to expect high consistency.
2. It is more promising to check (and potentially improve) label consistency on discoloration defects than scratch defects. Since HLP is lower on discoloration, it’s possible that there might be ambiguous labelling instructions that is affecting HLP.

Q6. To implement the data iteration loop effectively, the key is to take all the time that’s needed to construct the right dataset first, so that all development can be done on that dataset without needing to spend time to update the data.

1. False
2. True

Q7. You have a data pipeline for product recommendations that (i) cleans data by removing duplicate entries and spam, (ii) makes predictions. An engineering team improves the system used for step (i). If the trained model for step (ii) remains the same, what can we confidently conclude about the performance of the overall system?

1 point

1. It will definitely improve since the data is now more clean.
2. It’s not possible to say – it may perform better or worse.
3. It will get worse because stage (ii) is now experiencing data/concept drift.
4. It will get worse because changing an earlier stage in a data pipeline always results in worse performance of the later stages.

Q8. What is the primary goal of building a PoC (proof of concept) system?

1 point

1. To build a robust deployment system.
2. To collect sufficient data to build a robusts system for deployment.
3. To select the most appropriate ML architecture for a task.
4. To check feasibility and help decide if an application is workable and worth deploying.

Q9. MLOps tools can store meta-data to keep track of data provenance and lineage. What do the terms data provenance and lineage mean?

1. Data provenance refers to where the data comes from, and data lineage the sequence of processing steps applied to it.
2. Data provenance refers the input x, and data lineage refers to the output y.
3. Data provenance refers to the sequence of processing steps applied to a dataset, and data lineage refers to where the data comes from.
4. Data provenance refers data pipeline, and data lineage refers to the age of the data (i.e., how recently was it collected).

Q10. You are working on phone visual inspection, where the task is to use an input image, x, to classify defects, y. You have stored meta-data for your entire ML system, such as which factory each image came from. Which of the following are reasonable uses of meta-data?

1. As an alternative to having to comment your code.
2. To suggest tags or to generate insights during error analysis.
3. As another input provided to human labelers (in addition to the image x) to boost HLP.
4. Keeping track of data provenance and lineage.

Recap:

I hope this article would be helpful for you to find all the coursera Quiz Answers.
If this article helped you learn something new for free, let others know about this,
Want to learn more? Check out another course.

Enroll on Coursera

<<<Check out another Coursera Quiz Answers >>>

Interpersonal Communication For Engineering Leaders Coursera Quiz Answers

Industrial Biotechnology Coursera Quiz Answers

Functional Programming Principles In Scala Coursera Quiz Answers

Exploratory Data Analysis Coursera Quiz Answers

Get More Solutions :

Leetcode Solutions

Linkedinn Assessments Solutions