AI For Beginners 3 — Questions and Answers
Question 1: What is a 'hyperparameter' in the context of machine learning?
- A parameter learned from training data
- A configuration set before training that controls the learning process (Correct answer)
- A weight in the neural network
- An output produced by the model
Correct answer: A configuration set before training that controls the learning process
Hyperparameters like learning rate and layer count are set by the engineer before training, not learned from data.
Question 2: Which technique randomly drops neurons during training to reduce overfitting?
- Batch normalization
- Dropout (Correct answer)
- Weight decay
- Data augmentation
Correct answer: Dropout
Dropout randomly deactivates a fraction of neurons during each training step, forcing the network to learn redundant representations.
Question 3: What does 'transfer learning' allow an AI engineer to do?
- Train a model from scratch on a new dataset
- Reuse a pre-trained model's knowledge for a new but related task (Correct answer)
- Transfer data between two databases
- Move a model from one programming language to another
Correct answer: Reuse a pre-trained model's knowledge for a new but related task
Transfer learning leverages representations learned on a large dataset (e.g., ImageNet) to accelerate training on a smaller target task.
Question 4: In a confusion matrix for binary classification, what does a 'false positive' represent?
- The model correctly predicted the positive class
- The model predicted positive but the true label is negative (Correct answer)
- The model predicted negative but the true label is positive
- The model correctly predicted the negative class
Correct answer: The model predicted positive but the true label is negative
A false positive (Type I error) occurs when the model incorrectly labels a negative example as positive.
Question 5: What is the role of the learning rate in gradient descent?
- It determines the number of training epochs
- It controls how large a step is taken in the direction of the gradient (Correct answer)
- It sets the number of neurons per layer
- It defines the loss function shape
Correct answer: It controls how large a step is taken in the direction of the gradient
The learning rate scales the gradient update; too high causes divergence, too low causes very slow convergence.
Question 6: Which of the following best describes an embedding in AI?
- A rule-based lookup table
- A dense vector representation of data in a continuous space (Correct answer)
- A one-hot encoded categorical variable
- A compressed image file format
Correct answer: A dense vector representation of data in a continuous space
Embeddings map discrete objects (words, items) to dense vectors so that similar objects are close together in vector space.
Question 7: What is 'data augmentation' primarily used for in AI training?
- Cleaning corrupted records from the dataset
- Artificially expanding the training set by applying transformations to existing samples (Correct answer)
- Removing duplicate entries from a database
- Increasing the model's parameter count
Correct answer: Artificially expanding the training set by applying transformations to existing samples
Data augmentation applies transforms like flipping, cropping, or noise addition to create more varied training examples from existing data.
AI For Beginners 3
What is a 'hyperparameter' in the context of machine learning?