Font size
Worksheets02_CS_3 Final Exam
Total questions: 90
Worksheet time: 45mins
Which of the following is a common method for collecting data to train AI models?
Data mining
Data painting
Data erasing
Data hiding
What is the main purpose of data preprocessing in AI model training?
To delete all data
To clean and organize data
To make data more confusing
To hide data from the model
Which algorithm is commonly used for training supervised learning models?
K-means clustering
Linear regression
Random search
Genetic algorithm
What does the term "overfitting" refer to in AI model training?
Model fits training data too closely
Model ignores training data
Model fits test data perfectly
Model never learns
Transfer learning involves:
Training a model from scratch
Using a pre-trained model for a new task
Deleting all previous model weights
Ignoring previous knowledge
Which of the following is NOT a data collection method for AI models?
Web scraping
Surveys
Data augmentation
Sensor data
Which technique is used to handle missing values in a dataset?
Data duplication
Data imputation
Data hiding
Data deletion
Which algorithm is typically used for unsupervised learning?
Decision tree
K-means clustering
Logistic regression
Support vector machine
Underfitting occurs when a model:
Learns the training data too well
Fails to capture the underlying pattern
Has too many parameters
Is trained for too long
Which of the following is a benefit of transfer learning?
Requires more data
Reduces training time
Increases model complexity
Ignores previous knowledge
If you have a dataset with both images and text, which data collection method would be most appropriate?
Only web scraping
Combining multiple sources
Only surveys
Only sensor data
Why is normalization important in data preprocessing?
It makes all data values zero
It scales features to a similar range
It removes all outliers
It duplicates data
How does gradient descent help in training AI models?
By increasing the loss function
By minimizing the loss function
By deleting data points
By randomizing model weights
Which scenario is most likely to cause overfitting?
Using a very simple model
Training with too little data
Training with too much data
Using a very complex model
How can transfer learning be applied to a new image classification task?
Use a pre-trained model and fine-tune it on new images
Train a new model from scratch
Ignore all previous models
Use only text data
Which data collection method would you use to gather real-time weather data for an AI model?
Surveys
Sensor data
Web scraping
Data augmentation
What is the effect of removing outliers during data preprocessing?
Increases noise in the data
Improves model accuracy
Makes data less reliable
Reduces data size only
Which model training algorithm is best suited for classification tasks?
Linear regression
Logistic regression
K-means clustering
Principal component analysis
How can you detect underfitting in a trained AI model?
High accuracy on training data, low on test data
Low accuracy on both training and test data
High accuracy on test data only
High accuracy on training data only
Which step is essential when applying transfer learning to a new domain?
Ignore the pre-trained model
Fine-tune the pre-trained model on new data
Use the pre-trained model without any changes
Delete all previous weights
You are tasked with building an AI model to predict house prices. How would you ensure the data collected is representative of the population?
Collect data from only one neighborhood
Collect data from diverse neighborhoods and house types
Use only old data
Ignore location information
Given a dataset with missing values, which preprocessing strategy would you use to minimize bias in your AI model?
Remove all rows with missing values
Impute missing values using the mean or median
Fill missing values with zeros
Ignore missing values
You are training a neural network and notice that the loss function is not decreasing. What could be a possible reason?
Learning rate is too high or too low
Model is overfitting
Data is perfectly clean
Model is too simple
How would you address overfitting in a deep learning model?
Add more layers to the model
Use regularization techniques like dropout
Reduce the amount of training data
Increase the learning rate
You want to use transfer learning for a medical image classification task. What is the best approach?
Use a pre-trained model on natural images and fine-tune with medical images
Train a model from scratch with medical images only
Use a pre-trained model without any changes
Use only text data
If your AI model performs well on training data but poorly on test data, what does this indicate and how would you fix it?
Overfitting; use techniques like cross-validation and regularization
Underfitting; add more features
Perfect model; do nothing
Data is too clean; add noise
You are given a large dataset with imbalanced classes. What preprocessing technique would you use to improve model performance?
Ignore the imbalance
Use resampling methods like oversampling or undersampling
Remove the majority class
Only use the minority class
How would you select the best model training algorithm for a given task?
Choose randomly
Analyze the problem type and data characteristics
Use the most popular algorithm
Use the fastest algorithm
You want to apply transfer learning to a text classification problem. What should you consider when choosing a pre-trained model?
The pre-trained model should be trained on similar text data
Any pre-trained model will work
Use a model trained on images
Ignore the domain of the pre-trained model
Suppose your AI model is underfitting. What strategy would you use to improve its performance?
Increase model complexity or add more features
Reduce the amount of training data
Use less data preprocessing
Lower the learning rate
What is the primary function of the Finch Robot?
To play music
To serve as an educational tool for learning programming and robotics
To clean floors
To cook food
Which programming language is commonly used to program the Finch Robot?
Python
HTML
CSS
SQL
Which sensor allows the Finch Robot to detect obstacles?
Temperature sensor
Light sensor
Distance sensor
Sound sensor
What is one application of the Finch Robot in education?
Teaching advanced calculus
Demonstrating basic programming concepts
Painting pictures
Cooking recipes
Which design principle is emphasized in the Finch Robot’s construction?
Complexity
Durability
Simplicity and accessibility
Luxury
What is the main purpose of the Finch Robot’s LED lights?
To provide entertainment
To indicate status or feedback
To heat the robot
To power the robot
Which sensor on the Finch Robot can measure ambient light?
Temperature sensor
Light sensor
Distance sensor
Pressure sensor
What is the typical power source for the Finch Robot?
Solar panels
Batteries
Gasoline
Wind energy
Which of the following is NOT a sensor found on the Finch Robot?
Temperature sensor
Light sensor
Pressure sensor
Distance sensor
What is the main goal of programming the Finch Robot in a classroom setting?
To win competitions
To learn coding and robotics concepts
To make phone calls
To browse the internet
If you want the Finch Robot to move forward for 2 seconds, which programming concept would you use?
Loop
Function call
Variable assignment
Conditional statement
How can you use the Finch Robot’s sensors to avoid obstacles?
By turning off the sensors
By programming the robot to change direction when the distance sensor detects an object
By ignoring sensor data
By moving randomly
Which programming structure would you use to make the Finch Robot repeat a movement pattern?
If statement
Loop
Print statement
Variable
How can the Finch Robot’s LED lights be used in a classroom project?
To display different colors based on sensor input
To cook food
To play music
To print documents
Which sensor would you use to program the Finch Robot to react to changes in room temperature?
Light sensor
Distance sensor
Temperature sensor
Sound sensor
How can students use the Finch Robot to learn about algorithms?
By memorizing robot parts
By writing code that controls the robot’s movements and responses
By drawing pictures of the robot
By reading stories about robots
If the Finch Robot’s light sensor detects a dark environment, what could you program it to do?
Move backward
Turn on its LED lights
Play a sound
Stop moving
How does the Finch Robot’s design support collaborative learning?
By being too complex for group work
By allowing multiple students to program and test it together
By limiting access to one user
By not supporting programming
Which sensor combination could be used to make the Finch Robot follow a line on the floor?
Temperature and sound sensors
Light and distance sensors
Pressure and humidity sensors
GPS and camera sensors
How can the Finch Robot be used to demonstrate the concept of feedback in a system?
By ignoring sensor data
By using sensor input to adjust its actions, such as changing direction when an obstacle is detected
By moving randomly
By turning off all sensors
A student wants the Finch Robot to avoid obstacles and reach a target location. What steps should they take to plan their program?
Write random code and hope for the best
Identify sensors needed, plan movement logic, and test the program iteratively
Ignore sensor data and move straight
Only use LED lights
How can you use evidence from sensor readings to improve the Finch Robot’s performance in a maze?
Ignore sensor readings
Analyze sensor data to adjust movement and avoid walls more efficiently
Only use pre-written code
Move randomly
A teacher wants students to use the Finch Robot to model a real-world system. Which approach demonstrates strategic thinking?
Program the robot to move in a straight line
Design a simulation where the robot mimics traffic flow using sensors and programmed rules
Only use the robot’s LED lights
Ignore sensor input
How can students use the Finch Robot to investigate the relationship between light intensity and robot movement?
Move the robot randomly
Collect light sensor data and program the robot to change speed based on light levels, then analyze results
Only use temperature sensor
Ignore sensor data
A group of students wants to design a classroom activity using the Finch Robot to teach about environmental monitoring. What should they consider in their planning?
Only use the robot’s movement features
Select relevant sensors, define data collection methods, and plan how to interpret results
Ignore sensor capabilities
Use the robot for entertainment only
How can the Finch Robot’s design principles be applied to create a new educational robot?
Make the robot as complex as possible
Focus on simplicity, accessibility, and ease of programming for students
Use expensive materials
Ignore user needs
A student observes that the Finch Robot is not responding to temperature changes. What reasoning process should they use to troubleshoot?
Ignore the problem
Check the temperature sensor, review the code, and test with different temperature sources
Only change the batteries
Move the robot to a different room
How can the Finch Robot be used to demonstrate the concept of conditional logic in programming?
By moving in a straight line regardless of input
By using sensor data to make decisions, such as turning when an obstacle is detected
By only using LED lights
By ignoring all sensor input
A teacher wants students to compare the effectiveness of different sensor combinations for a specific task. What should students do?
Use only one sensor
Test various sensor combinations, collect data, and analyze which combination works best for the task
Ignore sensor data
Only use pre-written code
How can students use the Finch Robot to develop teamwork and problem-solving skills?
Work individually without sharing ideas
Collaborate to plan, program, and troubleshoot the robot’s tasks, reflecting on their strategies and outcomes
Only use the robot for entertainment
Ignore programming challenges
Which of the following is a common method for collecting data to train AI models?
Data mining
Data painting
Data erasing
Data hiding
What is the main purpose of data preprocessing in AI model training?
To delete all data
To clean and organize data
To make data more confusing
To hide data from the model
Which algorithm is commonly used for training supervised learning models?
K-means clustering
Linear regression
Random search
Genetic algorithm
What does the term "overfitting" refer to in AI model training?
Model fits training data too closely
Model ignores training data
Model fits test data perfectly
Model never learns
Transfer learning involves:
Training a model from scratch
Using a pre-trained model for a new task
Deleting all previous model weights
Ignoring previous knowledge
Which of the following is NOT a data collection method for AI models?
Web scraping
Surveys
Data augmentation
Sensor data
Which technique is used to handle missing values in a dataset?
Data duplication
Data imputation
Data hiding
Data deletion
Which algorithm is typically used for unsupervised learning?
Decision tree
K-means clustering
Logistic regression
Support vector machine
Underfitting occurs when a model:
Learns the training data too well
Fails to capture the underlying pattern
Has too many parameters
Is trained for too long
Which of the following is a benefit of transfer learning?
Requires more data
Reduces training time
Increases model complexity
Ignores previous knowledge
If you have a dataset with both images and text, which data collection method would be most appropriate?
Only web scraping
Combining multiple sources
Only surveys
Only sensor data
Why is normalization important in data preprocessing?
It makes all data values zero
It scales features to a similar range
It removes all outliers
It duplicates data
How does gradient descent help in training AI models?
By increasing the loss function
By minimizing the loss function
By deleting data points
By randomizing model weights
Which scenario is most likely to cause overfitting?
Using a very simple model
Training with too little data
Training with too much data
Using a very complex model
How can transfer learning be applied to a new image classification task?
Use a pre-trained model and fine-tune it on new images
Train a new model from scratch
Ignore all previous models
Use only text data
Which data collection method would you use to gather real-time weather data for an AI model?
Surveys
Sensor data
Web scraping
Data augmentation
What is the effect of removing outliers during data preprocessing?
Increases noise in the data
Improves model accuracy
Makes data less reliable
Reduces data size only
Which model training algorithm is best suited for classification tasks?
Linear regression
Logistic regression
K-means clustering
Principal component analysis
How can you detect underfitting in a trained AI model?
High accuracy on training data, low on test data
Low accuracy on both training and test data
High accuracy on test data only
High accuracy on training data only
Which step is essential when applying transfer learning to a new domain?
Ignore the pre-trained model
Fine-tune the pre-trained model on new data
Use the pre-trained model without any changes
Delete all previous weights
You are tasked with building an AI model to predict house prices. How would you ensure the data collected is representative of the population?
Collect data from only one neighborhood
Collect data from diverse neighborhoods and house types
Use only old data
Ignore location information
Given a dataset with missing values, which preprocessing strategy would you use to minimize bias in your AI model?
Remove all rows with missing values
Impute missing values using the mean or median
Fill missing values with zeros
Ignore missing values
You are training a neural network and notice that the loss function is not decreasing. What could be a possible reason?
Learning rate is too high or too low
Model is overfitting
Data is perfectly clean
Model is too simple
How would you address overfitting in a deep learning model?
Add more layers to the model
Use regularization techniques like dropout
Reduce the amount of training data
Increase the learning rate
You want to use transfer learning for a medical image classification task. What is the best approach?
Use a pre-trained model on natural images and fine-tune with medical images
Train a model from scratch with medical images only
Use a pre-trained model without any changes
Use only text data
If your AI model performs well on training data but poorly on test data, what does this indicate and how would you fix it?
Overfitting; use techniques like cross-validation and regularization
Underfitting; add more features
Perfect model; do nothing
Data is too clean; add noise
You are given a large dataset with imbalanced classes. What preprocessing technique would you use to improve model performance?
Ignore the imbalance
Use resampling methods like oversampling or undersampling
Remove the majority class
Only use the minority class
How would you select the best model training algorithm for a given task?
Choose randomly
Analyze the problem type and data characteristics
Use the most popular algorithm
Use the fastest algorithm
You want to apply transfer learning to a text classification problem. What should you consider when choosing a pre-trained model?
The pre-trained model should be trained on similar text data
Any pre-trained model will work
Use a model trained on images
Ignore the domain of the pre-trained model
Suppose your AI model is underfitting. What strategy would you use to improve its performance?
Increase model complexity or add more features
Reduce the amount of training data
Use less data preprocessing
Lower the learning rate
