Artificial Intelligence Artificial Intelligence 4 — Questions and Answers
Question 1: In a confusion matrix, what does a false positive represent?
- A negative case incorrectly predicted as positive (Correct answer)
- A positive case correctly predicted
- A positive case incorrectly predicted as negative
- A negative case correctly predicted
Correct answer: A negative case incorrectly predicted as positive
A false positive occurs when the model predicts the positive class for an instance that is actually negative.
Question 2: Which architecture introduced the self-attention mechanism that powers modern large language models?
- Recurrent neural network
- Restricted Boltzmann machine
- Transformer (Correct answer)
- Perceptron
Correct answer: Transformer
The 2017 Transformer architecture replaced recurrence with self-attention, enabling models like GPT and BERT.
Question 3: A self-driving car deciding when to brake based on sensor input is an example of which AI application area?
- Natural language processing
- Recommendation systems
- Speech synthesis
- Autonomous systems and perception (Correct answer)
Correct answer: Autonomous systems and perception
Autonomous vehicles combine perception and real-time decision-making, a hallmark of autonomous systems.
Question 4: What is the role of the learning rate in gradient descent?
- It sets the number of hidden layers
- It controls the size of each weight update step (Correct answer)
- It determines the dataset size
- It selects the activation function
Correct answer: It controls the size of each weight update step
The learning rate scales the gradient to determine how far weights move at each optimization step.
Question 5: Which of these is a generative AI task rather than a discriminative one?
- Classifying an email as spam
- Predicting whether a loan will default
- Creating a new image from a text description (Correct answer)
- Detecting faces in a photo
Correct answer: Creating a new image from a text description
Generative models produce new content, while discriminative models classify or predict labels for existing inputs.
Question 6: In knowledge representation, what does a semantic network use to encode information?
- Only numerical matrices
- Nodes for concepts and edges for relationships (Correct answer)
- Unstructured text documents
- Binary decision stumps
Correct answer: Nodes for concepts and edges for relationships
Semantic networks represent knowledge as a graph of concept nodes linked by labeled relationships.
Question 7: What does precision measure in a classification model?
- The fraction of actual positives that were found
- The total accuracy across all classes
- The speed of inference
- The fraction of positive predictions that were correct (Correct answer)
Correct answer: The fraction of positive predictions that were correct
Precision is true positives divided by all positive predictions, measuring prediction correctness.
Artificial Intelligence Artificial Intelligence 4
In a confusion matrix, what does a false positive represent?