Unit 2: AI Project Cycle

WORKSHEET XI AI

September 17, 2024

Unit 4: Data Sciences

September 19, 2024

109

AI Project Cycle

MCQs:

Which of the following is the first step in the AI Project Cycle?
A) Data Acquisition
B) Problem Scoping
C) Model Evaluation
D) Data Exploration
Answer: B
What is the purpose of the AI Project Cycle?
A) To create a project using hardware components
B) To follow a structured framework to develop AI solutions
C) To evaluate coding efficiency
D) To monitor project finances
Answer: B
Which of the following Sustainable Development Goals (SDGs) aim to end poverty?
A) Goal 1
B) Goal 5
C) Goal 7
D) Goal 9
Answer: A
What does the “Who” block in the 4Ws Problem Canvas focus on?
A) Identifying stakeholders affected by the problem
B) Finding locations for deploying the solution
C) Determining the project’s budget
D) Analyzing the technical requirements
Answer: A
Which of these describes “Training Data”?
A) Data used for testing the final model
B) Data used to evaluate project performance
C) Data fed to the machine to help it learn
D) Data collected after project deployment
Answer: C
What is “Testing Data”?
A) Data collected after the project is complete
B) Data used to verify the model’s accuracy
C) Data collected before training begins
D) Randomized data to confuse the model
Answer: B
Which of the following is NOT a method of Data Acquisition?
A) Surveys
B) Sensors
C) Modelling
D) Web Scraping
Answer: C
What is the purpose of visualizing data during Data Exploration?
A) To confuse the model
B) To understand patterns and trends in the data
C) To increase the data size
D) To reduce data authenticity
Answer: B
Which of the following is an example of a Rule-Based AI model?
A) Machine learning
B) A model that adapts to new data
C) A system based on fixed instructions
D) Deep learning
Answer: C
In which AI model does the machine learn without human labels on the dataset?
A) Supervised Learning
B) Unsupervised Learning
C) Reinforcement Learning
D) Rule-Based Model
Answer: B
What is the key advantage of Neural Networks over traditional machine learning models?
A) They require less data
B) They automatically extract data features
C) They are simpler to program
D) They do not require testing
Answer: B
What stage comes after “Problem Scoping” in the AI Project Cycle?
A) Model Evaluation
B) Data Acquisition
C) Data Exploration
D) Modelling
Answer: B
Which of the following is NOT one of the Sustainable Development Goals (SDGs)?
A) Quality Education
B) Climate Action
C) Space Exploration
D) Clean Water and Sanitation
Answer: C
In which type of learning does the machine get trained on a labeled dataset?
A) Unsupervised Learning
B) Reinforcement Learning
C) Supervised Learning
D) Deep Learning
Answer: C
What is the goal of the “Where” block in the 4Ws Problem Canvas?
A) Identify stakeholders
B) Analyze the problem context
C) Determine the project budget
D) Understand how AI models work
Answer: B
What does “Dimensionality Reduction” refer to in AI?
A) Reducing the number of features in the dataset
B) Increasing the complexity of the AI model
C) Visualizing data in 4D
D) Maximizing the number of data points
Answer: A
Which stage of the AI Project Cycle involves the use of graphs and charts?
A) Data Exploration
B) Problem Scoping
C) Model Evaluation
D) Data Acquisition
Answer: A
What is an example of Supervised Learning?
A) Clustering unlabelled data
B) Training a model with labeled images
C) Analyzing data trends without supervision
D) Reinforcing machine responses to mistakes
Answer: B
Which method is used to visualize data for patterns?
A) Web scraping
B) Surveys
C) Graphs and charts
D) Coding
Answer: C
What is the final stage of the AI Project Cycle?
A) Data Acquisition
B) Data Exploration
C) Model Evaluation
D) Problem Scoping
Answer: C
What is the “Why” block of the 4Ws Problem Canvas focused on?
A) Describing how the problem is important to stakeholders
B) Defining the technical requirements
C) Listing the project’s financial details
D) Visualizing the data model
Answer: A
What is “Clustering” in Unsupervised Learning?
A) Grouping data into labeled categories
B) Grouping similar data points based on patterns
C) Dividing data into training and testing sets
D) Reducing data dimensions
Answer: B
Which SDG focuses on gender equality?
A) Goal 1
B) Goal 5
C) Goal 10
D) Goal 13
Answer: B
What type of data is used in Regression models?
A) Randomized data
B) Labeled data
C) Continuous data
D) Unlabeled data
Answer: C
What is a key feature of a Learning-Based AI model?
A) It adapts to new data
B) It uses predefined rules
C) It does not require testing
D) It operates without data
Answer: A
What does a “Rule-Based Approach” in AI entail?
A) Creating models that adapt to changes
B) Pre-defining rules for the machine to follow
C) Allowing the machine to learn by itself
D) Reinforcing machine actions based on feedback
Answer: B
Which of the following is a key challenge with “Rule-Based” models?
A) Models improve over time
B) Models are dynamic and change frequently
C) Models cannot adapt to changes in data
D) Models automatically extract data features
Answer: C
What is the first task of a Neural Network’s input layer?
A) Processing the data
B) Acquiring data for the network
C) Filtering unnecessary data
D) Training the data
Answer: B
What is an example of Unsupervised Learning?
A) Training a machine with labeled data
B) Predicting future data trends
C) Discovering patterns from unlabelled data
D) Creating a rule-based system
Answer: C
What is one purpose of using a “Problem Statement Template”?
A) To collect more data
B) To analyze stakeholder feedback
C) To summarize the key elements of the problem
D) To test the final model
Answer: C
How can Dimensionality Reduction affect data?
A) It improves the quality of data
B) It adds more details to the dataset
C) It may cause loss of information
D) It increases data accuracy
Answer: C
Which of the following represents “Supervised Learning”?
A) Using unlabelled data for predictions
B) Learning from labeled training data
C) Creating a model without any human input
D) Clustering unknown data
Answer: B
What is a drawback of using Rule-Based AI models?
A) They adapt quickly to new data
B) They are dynamic in nature
C) They fail to improve with new data
D) They use labeled data for training
Answer: C
What is the final output of a Neural Network?
A) Hidden layer information
B) Input layer data
C) Processed data from the output layer
D) Unprocessed data
Answer: C
In which stage of the AI Project Cycle is the efficiency of the model calculated?
A) Problem Scoping
B) Model Evaluation
C) Data Acquisition
D) Modelling
Answer: B
Which of these is an example of Reinforcement Learning?
A) Training the model with labeled images
B) Adjusting actions based on rewards and punishments
C) Identifying clusters from unlabelled data
D) Visualizing trends in data
Answer: B
Which SDG focuses on Climate Action?
A) Goal 7
B) Goal 13
C) Goal 16
D) Goal 10
Answer: B
What type of data is used in Clustering?
A) Labelled data
B) Unlabelled data
C) Continuous data
D) Randomized data
Answer: B
What does the term “Data Features” refer to?
A) The labels assigned to data
B) The specific types of data collected
C) The amount of data required for training
D) The performance score of a model
Answer: B
Which of the following algorithms is used to reduce data complexity?
A) Clustering
B) Regression
C) Dimensionality Reduction
D) Reinforcement Learning
Answer: C
Which step helps in summarizing the “Problem Scoping” process?
A) Data Acquisition
B) Problem Statement Template
C) Clustering
D) Model Evaluation
Answer: B
What does “Exploratory Data Analysis” focus on?
A) Acquiring new data
B) Visualizing patterns and trends
C) Modelling the project
D) Creating new algorithms
Answer: B
What is “Precision” in the context of AI model evaluation?
A) The ratio of relevant instances retrieved
B) The ability to predict continuous values
C) The process of acquiring data features
D) The number of unlabelled data points
Answer: A
What is the role of Hidden Layers in Neural Networks?
A) To acquire data from the user
B) To process the input data
C) To display the final output
D) To define rules for predictions
Answer: B
In which AI model does the machine adjust actions based on feedback?
A) Supervised Learning
B) Unsupervised Learning
C) Reinforcement Learning
D) Rule-Based Learning
Answer: C
Which stage involves collecting relevant data for the AI model?
A) Data Acquisition
B) Model Evaluation
C) Problem Scoping
D) Data Exploration
Answer: A
What does “Recall” measure in model evaluation?
A) The amount of data correctly classified
B) The ability to identify all relevant cases
C) The process of reducing dimensions
D) The efficiency of training data
Answer: B
What does the “What” block in the 4Ws Problem Canvas help identify?
A) The stakeholders affected by the problem
B) The nature of the problem
C) The geographic location of the problem
D) The benefits of the solution
Answer: B
Which of the following best describes Supervised Learning?
A) Learning from unlabeled data
B) Learning from labeled data
C) Adjusting actions based on rewards
D) Visualizing patterns in 3D
Answer: B
What is the purpose of the output layer in Neural Networks?
A) To display results to the user
B) To process the input data
C) To adjust actions based on patterns
D) To cluster data into groups
Answer: A

Assertion (A): Problem Scoping is the first step in the AI Project Cycle.
Reason (R): Problem Scoping helps in identifying stakeholders and understanding the key elements of the problem.
Answer:
A) Both A and R are true, and R is the correct explanation of A.
B) Both A and R are true, but R is not the correct explanation of A.
C) A is true, but R is false.
D) A is false, but R is true.
Correct Answer: A
Assertion (A): Dimensionality reduction improves data accuracy.
Reason (R): Reducing dimensions simplifies the data but may lead to information loss.
Answer:
A) Both A and R are true, and R is the correct explanation of A.
B) Both A and R are true, but R is not the correct explanation of A.
C) A is true, but R is false.
D) A is false, but R is true.
Correct Answer: D
Assertion (A): Rule-based models are dynamic and can adapt to new data.
Reason (R): Rule-based models rely on predefined rules and cannot modify their behavior based on new data.
Answer:
A) Both A and R are true, and R is the correct explanation of A.
B) Both A and R are true, but R is not the correct explanation of A.
C) A is true, but R is false.
D) A is false, but R is true.
Correct Answer: D
Assertion (A): Neural networks tend to perform better with larger datasets.
Reason (R): Neural networks automatically extract data features, which helps them handle complex data patterns.
Answer:
A) Both A and R are true, and R is the correct explanation of A.
B) Both A and R are true, but R is not the correct explanation of A.
C) A is true, but R is false.
D) A is false, but R is true.
Correct Answer: A
Assertion (A): In supervised learning, the data used for training is unlabeled.
Reason (R): Supervised learning involves labeled datasets where the model is trained with input-output pairs.
Answer:
A) Both A and R are true, and R is the correct explanation of A.
B) Both A and R are true, but R is not the correct explanation of A.
C) A is true, but R is false.
D) A is false, but R is true.
Correct Answer: D

Case Study 1:

Case: A team of developers is building an AI model to predict house prices based on various factors such as location, size, number of rooms, and amenities. They gather a large dataset of historical house prices and begin exploring patterns in the data using visualizations like bar charts and scatter plots.

Question:
At which stage of the AI Project Cycle is the team currently working?
Answer:
The team is at the Data Exploration stage, where they are analyzing and visualizing the data to understand patterns and relationships.

Case Study 2:

Case: A company is using a supervised learning algorithm to classify customer feedback as either “positive” or “negative.” They have labeled data where feedback is already categorized, and they use this dataset to train the model. After training, they test the model with new feedback data that has not been labeled.

Question:
What kind of learning is being applied in this AI project, and what is the purpose of using new feedback data?
Answer:
This is a Supervised Learning approach. The new feedback data is used as Testing Data to evaluate the accuracy and performance of the model after training.

Case Study 3:

Case: A researcher is trying to build an AI system that can identify different dog breeds from images. The system is trained with a labeled dataset containing images of various dog breeds. However, when new breeds that were not in the training set are introduced, the system struggles to classify them correctly.

Question:
What might be the reason for the system’s failure, and what could be a possible solution?
Answer:
The failure occurs because the system is not able to generalize beyond the labeled training data, which is a limitation of Supervised Learning. A possible solution could be incorporating a Learning-Based Approach like Unsupervised Learning or increasing the diversity of the training dataset to cover more breeds.

Case Study 4:

Case: A healthcare organization is developing an AI model to predict the likelihood of patients developing diabetes based on historical health data. The team collects data from open-source government health databases and uses the information to train their model.

Question:
Which stage of the AI Project Cycle is the team in when they are gathering health data from databases, and why is this step important?
Answer:
The team is in the Data Acquisition stage. This step is important because gathering authentic, relevant data is crucial for training the AI model accurately and ensuring it makes reliable predictions.

Case Study 5:

Case: An AI company is developing a self-driving car system that uses a combination of cameras and sensors to navigate roads. Initially, the system follows pre-defined rules for actions such as turning and stopping at signals. Over time, the system adapts based on new driving scenarios and improves its decision-making ability.

Question:
What type of AI model is the company using initially, and how does it evolve over time?
Answer:
Initially, the company is using a Rule-Based Model, where pre-defined rules govern the system’s actions. Over time, it evolves into a Learning-Based Model, likely using Reinforcement Learning, as the system adapts and improves its decisions based on new driving scenarios.

QUESTION-ANSWERS:

1. What is the AI Project Cycle, and why is it important in AI development?

Answer:
The AI Project Cycle is a structured framework for developing AI projects, consisting of five main stages: Problem Scoping, Data Acquisition, Data Exploration, Modelling, and Evaluation. It is important because it provides a step-by-step guide to building efficient AI models by ensuring that all key elements such as problem identification, data collection, and evaluation are addressed systematically.

2. Explain the role of the “4Ws Problem Canvas” in Problem Scoping.

Answer:
The “4Ws Problem Canvas” helps in identifying key aspects of a problem by answering four questions:

Who is affected by the problem (stakeholders)?
What is the nature of the problem?
Where does the problem occur (context/situation)?
Why is solving the problem beneficial?
This helps in gaining a comprehensive understanding of the problem and its impact, leading to a clearer project goal.

3. What is the difference between Supervised and Unsupervised Learning in AI?

Answer:

Supervised Learning involves training the model with labeled data, where both the input and the expected output are known. It is used for tasks like classification and regression.
Unsupervised Learning works with unlabeled data, where the model identifies patterns and structures without explicit instructions. It is used for tasks like clustering and dimensionality reduction.

4. Why is Data Acquisition a critical stage in the AI Project Cycle, and what are the key challenges?

Answer:
Data Acquisition is critical because the quality and relevance of the data directly impact the accuracy of the AI model. If the data is incomplete, irrelevant, or biased, the model’s predictions will be flawed. Key challenges include finding authentic, relevant data sources and ensuring that the data is unbiased and legal to use, such as open-sourced or publicly available datasets.

5. Describe the concept of Dimensionality Reduction and its significance in AI.

Answer:
Dimensionality Reduction refers to reducing the number of features in a dataset while retaining its essential information. It is significant because it helps in simplifying data, reducing computational costs, and minimizing the risk of overfitting. However, reducing dimensions must be done carefully, as too much reduction can lead to information loss, negatively affecting the model’s performance.

6. What are Neural Networks, and how do they differ from traditional machine learning algorithms?

Answer:
Neural Networks are a type of AI model inspired by the human brain’s structure, consisting of layers of interconnected nodes (neurons). They are particularly effective for tasks like image and speech recognition due to their ability to automatically extract features from large datasets. Unlike traditional machine learning models, which rely on manually defined features, neural networks can learn complex patterns without extensive human input.

7. Explain the concept of “Rule-Based AI” and its limitations.

Answer:
Rule-Based AI models operate on pre-defined rules set by the developer. The machine follows these instructions to perform tasks. While this approach is simple and easy to implement, its limitation is that it cannot adapt to new or changing data. Once trained, it remains static and cannot learn or improve based on new information or feedback.

8. What is the importance of Data Exploration in the AI Project Cycle?

Answer:
Data Exploration is essential because it involves visualizing and analyzing data to understand patterns, relationships, and trends. This helps in identifying key data features and informs the selection of the appropriate AI model. By visualizing data using graphs, charts, and other techniques, teams can better interpret the data and communicate insights more effectively.

9. How does Reinforcement Learning differ from Supervised Learning?

Answer:
In Reinforcement Learning, the model learns by interacting with its environment and receiving feedback in the form of rewards or punishments based on its actions. It is typically used in scenarios like game playing or robotics. In contrast, Supervised Learning involves training a model with labeled data, where the correct output is known beforehand, and the model is evaluated based on its ability to predict that output.

10. What is the significance of evaluating an AI model after it has been built?

Answer:
Evaluating an AI model is crucial to assess its performance, accuracy, and efficiency. This process involves testing the model on new data (Testing Data) and calculating metrics like accuracy, precision, recall, and F1 score. Proper evaluation helps in identifying areas for improvement and ensures that the model meets the project’s objectives before it is deployed for real-world use.

11. What are the benefits of using data visualization in the AI project cycle?

Answer:
Data visualization helps in:

Quickly identifying trends, patterns, and outliers in the data.
Communicating complex data insights to stakeholders in an easily understandable format.
Aiding in the selection of appropriate AI models based on observed patterns. It is a key tool in Data Exploration, ensuring that the data is interpreted accurately and used effectively in model development.

12. What challenges might arise during Problem Scoping in AI projects, and how can they be addressed?

Answer:
Challenges in Problem Scoping include:

Difficulty in identifying a well-defined problem.
Misunderstanding the needs of stakeholders.
Failing to capture the full context of the problem. These can be addressed by using frameworks like the “4Ws Problem Canvas” to clearly define the problem, gathering input from stakeholders, and conducting thorough research to understand the scope and impact of the problem.

13. Why is authentic data important in AI model training, and what risks are associated with poor data quality?

Answer:
Authentic data is important because it ensures the AI model is trained on accurate and reliable information, leading to better predictions and decisions. Poor data quality, such as incomplete, irrelevant, or biased data, can lead to incorrect or biased results, reduced model performance, and ethical concerns, such as reinforcing harmful stereotypes.

14. Describe the key differences between Classification and Regression in Supervised Learning.

Answer:
In Classification, the output variable is categorical, and the task is to classify data into predefined categories or classes (e.g., classifying emails as “spam” or “not spam”).
In Regression, the output variable is continuous, and the task is to predict a numerical value based on input data (e.g., predicting house prices based on features like size and location).

15. What are the main performance metrics used in evaluating an AI model, and what do they measure?

Answer:
Key performance metrics include:

Accuracy: The proportion of correct predictions out of all predictions.
Precision: The proportion of true positive predictions out of all positive predictions.
Recall: The proportion of true positive predictions out of all actual positives.
F1 Score: The harmonic mean of precision and recall, providing a balance between the two. These metrics help evaluate the model’s effectiveness and identify areas for improvement.

ai cbse

This site is dedicated to provide contents, notes, questions bank,blogs,articles and other materials for AI students of CBSE.