Florida AI & Machine Learning Specialist
Exam Practice Questions And Correct
Answers (Verified Answers) Plus Rationales
2025|2026 Q&A | Instant Download Pdf
1. What is the primary goal of supervised learning?
To learn a mapping from inputs to outputs using labeled data
Supervised learning uses labeled datasets where input-output pairs are
known, allowing the model to learn a function that generalizes to
unseen data.
2. What is the key difference between classification and regression?
Classification predicts discrete labels, while regression predicts
continuous values
Classification assigns data into categories, while regression estimates
numerical values.
3. In machine learning, what does overfitting mean?
The model performs well on training data but poorly on unseen data
, Overfitting occurs when a model memorizes training data instead of
generalizing patterns.
4. What is underfitting?
The model fails to capture patterns in the data and performs poorly
on both training and test sets
Underfitting happens when the model is too simple to learn the
underlying data distribution.
5. What is the purpose of cross-validation?
To assess model performance by splitting data into multiple train-
test folds
Cross-validation provides a more reliable estimate of generalization
performance than a single train-test split.
6. What does feature scaling do?
It standardizes or normalizes input variables to improve model
convergence
Many algorithms, like gradient descent, perform better when features
are on the same scale.
7. What is gradient descent used for in machine learning?
To optimize model parameters by minimizing a loss function
Gradient descent iteratively updates weights in the direction that
reduces error.
,8. What is the difference between batch and stochastic gradient
descent?
Batch uses the entire dataset per update, while stochastic uses one
sample at a time
Batch descent is stable but slow, while stochastic descent is faster but
noisier.
9. What is the role of a loss function in training a machine learning
model?
It measures how far predictions are from actual values
The loss guides optimization by quantifying prediction errors.
10. What is regularization in machine learning?
A technique to reduce overfitting by penalizing large weights
Regularization methods like L1 (Lasso) and L2 (Ridge) constrain model
complexity.
11. What is the purpose of an activation function in neural
networks?
To introduce non-linearity into the model
Without activation functions, neural networks behave like linear
models and cannot capture complex patterns.
12. What is the vanishing gradient problem?
When gradients become too small during backpropagation, slowing
learning
, This often happens with deep networks using sigmoid or tanh
activations, causing ineffective weight updates.
13. What is the exploding gradient problem?
When gradients grow excessively large during training, leading to
unstable updates
Exploding gradients can cause numerical instability and poor
convergence.
14. What is the main advantage of using ReLU as an activation
function?
It reduces vanishing gradients and allows faster training
ReLU outputs zero for negatives and linear for positives, keeping
gradients stable.
15. What is dropout in neural networks?
A regularization method that randomly disables neurons during
training
Dropout prevents overfitting by reducing co-dependence among
neurons.
16. What is early stopping?
A technique to halt training when validation error stops improving
Early stopping prevents overfitting by not overtraining the model.
17. What is transfer learning?
Using a pre-trained model on one task as a starting point for another
Exam Practice Questions And Correct
Answers (Verified Answers) Plus Rationales
2025|2026 Q&A | Instant Download Pdf
1. What is the primary goal of supervised learning?
To learn a mapping from inputs to outputs using labeled data
Supervised learning uses labeled datasets where input-output pairs are
known, allowing the model to learn a function that generalizes to
unseen data.
2. What is the key difference between classification and regression?
Classification predicts discrete labels, while regression predicts
continuous values
Classification assigns data into categories, while regression estimates
numerical values.
3. In machine learning, what does overfitting mean?
The model performs well on training data but poorly on unseen data
, Overfitting occurs when a model memorizes training data instead of
generalizing patterns.
4. What is underfitting?
The model fails to capture patterns in the data and performs poorly
on both training and test sets
Underfitting happens when the model is too simple to learn the
underlying data distribution.
5. What is the purpose of cross-validation?
To assess model performance by splitting data into multiple train-
test folds
Cross-validation provides a more reliable estimate of generalization
performance than a single train-test split.
6. What does feature scaling do?
It standardizes or normalizes input variables to improve model
convergence
Many algorithms, like gradient descent, perform better when features
are on the same scale.
7. What is gradient descent used for in machine learning?
To optimize model parameters by minimizing a loss function
Gradient descent iteratively updates weights in the direction that
reduces error.
,8. What is the difference between batch and stochastic gradient
descent?
Batch uses the entire dataset per update, while stochastic uses one
sample at a time
Batch descent is stable but slow, while stochastic descent is faster but
noisier.
9. What is the role of a loss function in training a machine learning
model?
It measures how far predictions are from actual values
The loss guides optimization by quantifying prediction errors.
10. What is regularization in machine learning?
A technique to reduce overfitting by penalizing large weights
Regularization methods like L1 (Lasso) and L2 (Ridge) constrain model
complexity.
11. What is the purpose of an activation function in neural
networks?
To introduce non-linearity into the model
Without activation functions, neural networks behave like linear
models and cannot capture complex patterns.
12. What is the vanishing gradient problem?
When gradients become too small during backpropagation, slowing
learning
, This often happens with deep networks using sigmoid or tanh
activations, causing ineffective weight updates.
13. What is the exploding gradient problem?
When gradients grow excessively large during training, leading to
unstable updates
Exploding gradients can cause numerical instability and poor
convergence.
14. What is the main advantage of using ReLU as an activation
function?
It reduces vanishing gradients and allows faster training
ReLU outputs zero for negatives and linear for positives, keeping
gradients stable.
15. What is dropout in neural networks?
A regularization method that randomly disables neurons during
training
Dropout prevents overfitting by reducing co-dependence among
neurons.
16. What is early stopping?
A technique to halt training when validation error stops improving
Early stopping prevents overfitting by not overtraining the model.
17. What is transfer learning?
Using a pre-trained model on one task as a starting point for another
