Font size
WorksheetsDaily Quiz (02.01.2021)
Total questions: 16
Worksheet time: 9mins
Common issues with training data are _____
Unbalanced or biased data
Data with more features
Mislabeled data
Insufficient data
Data doesn't reflect the real world
What are the three steps of building AI products?
Identify the business problem and ideate a solution
Make viable business strategy
Prototype, test and refine
Release, measure and update
As a product manager, problems can come from ____
Pressure from investor
Upset users or customers
Lost revenue
Grumpy Engineers
Struggling sales and customer teams
When we talk about impact, we often want to think about the primary users
True
False
Is this the real prototype cycle?
Yes
No
Often, when you are building a product, you'll want to measure how effective your product is at its overall goal. In this video annotation case, the goal was to make it easier/automatic for users to annotate video frames, and this efficacy was measured in a specific way.
How could efficacy be measured for this video annotation product?
The time spent on Figure Eight's platform
The number of human-generated annotations vs. machine-generated
User engagement and rating of the video annotation tool
Efficacy is indicated by how many times a human had to annotate a video frame (hopefully very few times) vs. how many times the frame was automatically annotated.
False
True
Which method we should use in case of annotating very very small object in video annotating task?
Computer vision
Image detection algorithm
Linear interpolation
Convolutional neural network
Bias issues are less in face recognition system.
False
True
Unbalanced selection of sorts data generate _____
Selection bias
Data bias
Model bias
Annotation bias
Suppose you are building a voice assistance and data is collected from same and less different race, same age groups and gender, then your model may perform better in real world.
True
False
A model was given a large volume of training data: unlabeled text, taken from Google News. The text itself is reflective of Google users and journalists
What kind of bias is responsible for the model learning these problematic word relationships?
Data bias
Model bias
Annotation bias
Data and model bias
Individual bias
What are the steps to solve unwanted bias?
Awareness
Resource engagement
Data management
Iteration and learning
Say you are developing a customer service chatbot for an airline, you've trained it to manage flight switches and cancellations, but it struggles to help customers with lost-luggage cases.
What kind of training data would you look for to improve your model?
Chatbot answers to flight accommodation requests
Chatbot queries that are all labeled as lost luggage cases
A variety of chatbot queries that are labeled by case: cancellation request, switching flights, and lost luggage
If you are aiming for high precision, which of these categories should have a high count?
True positives
True negative
False positives
False negative
______ are related because you are dealing with huge volumes of data
Storage challenges
Network challenges
Compute challenges
