LATEST!!! ARTIBA Artificial Intelligence
Engineer Certification – Practice Exam
Questions Verified Answers With Rationale
And Complete Solutions Grade A+.
Duration: 90 minutes
Total Questions:150
Instructions: Choose the best answer for multiple-choice questions. For scenario-based
questions, provide concise, accurate answers.
Section A: Machine Learning Fundamentals (15
Questions)
Q1. Which of the following is an example of unsupervised learning?
A) Linear regression
B) Logistic regression
C) K-means clustering
D) Random forest
Answer: C) K-means clustering
Rationale: K-means clustering groups data without labeled outputs, which is
the definition of unsupervised learning.
Q2. You are training a model and notice it performs extremely well on training data but
poorly on validation data. What is this called?
A) Underfitting
B) Overfitting
C) Data leakage
D) Regularization
Answer: B) Overfitting
Rationale: Overfitting occurs when a model memorizes training data but fails to
generalize to unseen data.
Page | 1
,Q3. Which of the following metrics is most suitable for evaluating a classification model
with imbalanced classes?
A) Accuracy
B) Precision and recall
C) Mean squared error
D) R-squared
Answer: B) Precision and recall
Rationale: For imbalanced datasets, accuracy can be misleading; precision and
recall better capture model performance on minority classes.
Q4. What is the main purpose of the activation function in a neural network?
A) Initialize weights
B) Introduce non-linearity
C) Compute loss
D) Reduce overfitting
Answer: B) Introduce non-linearity
Rationale: Activation functions allow neural networks to model complex, non-
linear relationships.
Q5. Which of the following is a regularization technique to prevent overfitting in deep
learning?
A) Dropout
B) Gradient descent
C) PCA
D) Hyperparameter tuning
Answer: A) Dropout
Rationale: Dropout randomly deactivates neurons during training, helping the
network generalize better.
Q6. What type of machine learning algorithm is typically used for predicting continuous
values?
A) Classification
B) Regression
C) Clustering
D) Reinforcement
Answer: B) Regression
Rationale: Regression models predict numerical outputs.
Page | 2
,Q7. Which of these is a common distance metric used in clustering algorithms?
A) Cross-entropy
B) Euclidean distance
C) ReLU
D) Softmax
Answer: B) Euclidean distance
Rationale: Euclidean distance measures similarity between points in space for
clustering.
Q8. In reinforcement learning, the agent learns primarily through:
A) Labeled datasets
B) Rewards and penalties
C) Unsupervised clustering
D) Backpropagation
Answer: B) Rewards and penalties
Rationale: Reinforcement learning relies on trial-and-error interactions with
the environment.
Q9. Which of the following is true about gradient descent?
A) It always finds the global minimum
B) It updates weights in the direction opposite to the gradient
C) It is used only in classification tasks
D) It cannot be used in deep learning
Answer: B) It updates weights in the direction opposite to the gradient
Rationale: Gradient descent iteratively updates weights to minimize loss by
moving against the gradient.
Q10. Which sampling technique helps balance imbalanced datasets?
A) Cross-validation
B) SMOTE (Synthetic Minority Oversampling Technique)
C) K-fold
D) PCA
Answer: B) SMOTE
Rationale: SMOTE generates synthetic samples of the minority class to balance
datasets.
Page | 3
, Q11. Which algorithm is most suitable for image recognition tasks?
A) Decision trees
B) Convolutional Neural Networks (CNNs)
C) KNN
D) Linear regression
Answer: B) Convolutional Neural Networks (CNNs)
Rationale: CNNs automatically detect spatial hierarchies in images.
Q12. Early stopping is a technique used to:
A) Accelerate training
B) Prevent overfitting
C) Normalize data
D) Increase batch size
Answer: B) Prevent overfitting
Rationale: Training stops when validation loss stops improving to avoid
overfitting.
Q13. Which of the following is NOT a type of ensemble learning?
A) Bagging
B) Boosting
C) Stacking
D) Normalization
Answer: D) Normalization
Rationale: Normalization scales data, not an ensemble technique.
Q14. Which loss function is commonly used for multi-class classification?
A) Mean squared error
B) Cross-entropy
C) Hinge loss
D) MAE
Answer: B) Cross-entropy
Rationale: Cross-entropy measures the difference between predicted probability
distributions and true labels.
Page | 4
Engineer Certification – Practice Exam
Questions Verified Answers With Rationale
And Complete Solutions Grade A+.
Duration: 90 minutes
Total Questions:150
Instructions: Choose the best answer for multiple-choice questions. For scenario-based
questions, provide concise, accurate answers.
Section A: Machine Learning Fundamentals (15
Questions)
Q1. Which of the following is an example of unsupervised learning?
A) Linear regression
B) Logistic regression
C) K-means clustering
D) Random forest
Answer: C) K-means clustering
Rationale: K-means clustering groups data without labeled outputs, which is
the definition of unsupervised learning.
Q2. You are training a model and notice it performs extremely well on training data but
poorly on validation data. What is this called?
A) Underfitting
B) Overfitting
C) Data leakage
D) Regularization
Answer: B) Overfitting
Rationale: Overfitting occurs when a model memorizes training data but fails to
generalize to unseen data.
Page | 1
,Q3. Which of the following metrics is most suitable for evaluating a classification model
with imbalanced classes?
A) Accuracy
B) Precision and recall
C) Mean squared error
D) R-squared
Answer: B) Precision and recall
Rationale: For imbalanced datasets, accuracy can be misleading; precision and
recall better capture model performance on minority classes.
Q4. What is the main purpose of the activation function in a neural network?
A) Initialize weights
B) Introduce non-linearity
C) Compute loss
D) Reduce overfitting
Answer: B) Introduce non-linearity
Rationale: Activation functions allow neural networks to model complex, non-
linear relationships.
Q5. Which of the following is a regularization technique to prevent overfitting in deep
learning?
A) Dropout
B) Gradient descent
C) PCA
D) Hyperparameter tuning
Answer: A) Dropout
Rationale: Dropout randomly deactivates neurons during training, helping the
network generalize better.
Q6. What type of machine learning algorithm is typically used for predicting continuous
values?
A) Classification
B) Regression
C) Clustering
D) Reinforcement
Answer: B) Regression
Rationale: Regression models predict numerical outputs.
Page | 2
,Q7. Which of these is a common distance metric used in clustering algorithms?
A) Cross-entropy
B) Euclidean distance
C) ReLU
D) Softmax
Answer: B) Euclidean distance
Rationale: Euclidean distance measures similarity between points in space for
clustering.
Q8. In reinforcement learning, the agent learns primarily through:
A) Labeled datasets
B) Rewards and penalties
C) Unsupervised clustering
D) Backpropagation
Answer: B) Rewards and penalties
Rationale: Reinforcement learning relies on trial-and-error interactions with
the environment.
Q9. Which of the following is true about gradient descent?
A) It always finds the global minimum
B) It updates weights in the direction opposite to the gradient
C) It is used only in classification tasks
D) It cannot be used in deep learning
Answer: B) It updates weights in the direction opposite to the gradient
Rationale: Gradient descent iteratively updates weights to minimize loss by
moving against the gradient.
Q10. Which sampling technique helps balance imbalanced datasets?
A) Cross-validation
B) SMOTE (Synthetic Minority Oversampling Technique)
C) K-fold
D) PCA
Answer: B) SMOTE
Rationale: SMOTE generates synthetic samples of the minority class to balance
datasets.
Page | 3
, Q11. Which algorithm is most suitable for image recognition tasks?
A) Decision trees
B) Convolutional Neural Networks (CNNs)
C) KNN
D) Linear regression
Answer: B) Convolutional Neural Networks (CNNs)
Rationale: CNNs automatically detect spatial hierarchies in images.
Q12. Early stopping is a technique used to:
A) Accelerate training
B) Prevent overfitting
C) Normalize data
D) Increase batch size
Answer: B) Prevent overfitting
Rationale: Training stops when validation loss stops improving to avoid
overfitting.
Q13. Which of the following is NOT a type of ensemble learning?
A) Bagging
B) Boosting
C) Stacking
D) Normalization
Answer: D) Normalization
Rationale: Normalization scales data, not an ensemble technique.
Q14. Which loss function is commonly used for multi-class classification?
A) Mean squared error
B) Cross-entropy
C) Hinge loss
D) MAE
Answer: B) Cross-entropy
Rationale: Cross-entropy measures the difference between predicted probability
distributions and true labels.
Page | 4
