AI0-001 Machine Learning and Deep Learning — Timed 50 Questions

Question 1

A data scientist is building a classification model to detect fraudulent transactions. The dataset is highly imbalanced with only 1% fraudulent cases. Which approach should the scientist use to evaluate model performance most effectively?

Accepted Answer

F1 score. In highly imbalanced datasets like fraud detection (1% positive class), accuracy is misleading because a model that predicts all transactions as legitimate would achieve 99% accuracy yet fail to detect any fraud. The F1 score (harmonic mean of precision and recall) is the most effective metric because it balances both false positives and false negatives, providing a single score that reflects the model's ability to correctly identify the minority class without being skewed by class imbalance.

Answer

Accuracy

Answer

Recall

Answer

Precision

Question 2

A machine learning team is deploying a model that predicts customer churn. They notice that the model's predictions are highly sensitive to small changes in input features, leading to inconsistent outputs. Which technique should the team apply to improve model stability?

Accepted Answer

Regularization. Regularization (Option C) is the correct technique because it adds a penalty term to the loss function (e.g., L1 or L2 regularization), which constrains the model's weights. This reduces variance and prevents overfitting to noise in the training data, directly addressing the high sensitivity to small input changes (brittleness). By shrinking coefficients, regularization forces the model to learn more general patterns, improving stability and consistency in predictions.

Answer

Increase learning rate

Answer

Feature scaling

Answer

Cross-validation

Question 3

A deep learning model for image classification is overfitting the training data. The team has already tried data augmentation and dropout. Which additional technique should they implement to reduce overfitting?

Accepted Answer

Early stopping. Early stopping (Option D) is the correct additional technique because it halts training when validation performance stops improving, directly preventing the model from memorizing noise in the training data. Since data augmentation and dropout are already in use, early stopping provides a complementary regularization effect by limiting the number of training iterations before overfitting occurs.

Answer

Batch normalization

Answer

Increase number of epochs

Answer

Gradient clipping

Question 4

A company wants to deploy a machine learning model that requires continuous learning as new data arrives. The model must be able to adapt to changing patterns without retraining from scratch. Which approach should be used?

Accepted Answer

Online learning. Online learning (also called incremental learning) updates the model incrementally as each new data point arrives, without requiring full retraining. This makes it ideal for scenarios where data arrives continuously and patterns shift over time, as the model can adapt its parameters on the fly.

Answer

Transfer learning

Answer

Batch learning

Answer

Unsupervised learning

Question 5

A data engineer is designing a pipeline to train a linear regression model on a dataset with 10 million rows and 50 features. The dataset fits in memory. Which approach should the engineer use to train the model efficiently?

Accepted Answer

Stochastic gradient descent. Stochastic gradient descent (SGD) is the most efficient approach for training a linear regression model on a dataset with 10 million rows and 50 features because it updates the model parameters using only one training example per iteration, leading to much faster convergence per epoch compared to batch methods. Since the dataset fits in memory, SGD can still be implemented efficiently without the overhead of loading data in batches from disk, and it scales well to large datasets where the normal equation or batch gradient descent would be computationally prohibitive.

Answer

Normal equation

Answer

Batch gradient descent

Answer

Principal component analysis

Question 6

A data scientist is training a convolutional neural network (CNN) for object detection. The training loss decreases rapidly but then plateaus at a high value, and the validation loss starts increasing. Which action should the scientist take to improve the model?

Accepted Answer

Reduce the model complexity. The training loss decreasing rapidly then plateauing at a high value while validation loss increases is classic overfitting. Reducing model complexity (Option C) directly addresses overfitting by decreasing the number of parameters or applying regularization (e.g., dropout, L2), which forces the network to learn more generalizable features rather than memorizing noise in the training data.

Answer

Increase the learning rate

Answer

Increase the number of epochs

Answer

Add more convolutional layers

Question 7

A team is building a recommendation system using collaborative filtering. They have a sparse user-item matrix. Which technique should they use to handle the sparsity and improve recommendations?

Accepted Answer

Matrix factorization. Matrix factorization (B) is the correct technique because it decomposes the sparse user-item matrix into lower-dimensional latent factor matrices, effectively capturing underlying patterns and filling in missing entries. This directly addresses sparsity by learning dense representations that generalize beyond observed interactions, which is a core strength in collaborative filtering for recommendation systems.

Answer

Association rule mining

Answer

k-nearest neighbors

Answer

Content-based filtering

Question 8

Which TWO techniques are commonly used to handle missing data in a machine learning dataset? (Choose TWO.)

Accepted Answer

Imputation with mean or median. Imputation with mean or median is a standard technique for handling missing numerical data because it preserves the dataset size and avoids introducing bias from simply discarding rows. By replacing missing values with the central tendency of the observed data, the model can still learn patterns without losing information, though it may reduce variance slightly.

Answer

Normalization

Answer

One-hot encoding

Answer

Dimensionality reduction

Question 9

Which THREE are common activation functions used in neural networks? (Choose THREE.)

Accepted Answer

ReLU. ReLU (Rectified Linear Unit) is a common activation function in neural networks because it introduces non-linearity while being computationally efficient. It outputs the input directly if positive, otherwise zero, which helps mitigate the vanishing gradient problem compared to sigmoid or tanh. This makes it a default choice for hidden layers in many deep learning architectures.

Answer

Softmax

Answer

Linear

Question 10

Which TWO are valid techniques to reduce overfitting in a deep neural network? (Choose TWO.)

Accepted Answer

L2 regularization. L2 regularization (option C) is a valid technique to reduce overfitting by adding a penalty term proportional to the square of the weight magnitudes to the loss function. This discourages the network from learning overly complex patterns, effectively shrinking weights and improving generalization. Dropout (option E) randomly drops a fraction of neurons during training, which prevents co-adaptation of features and forces the network to learn more robust representations, also reducing overfitting.

Answer

Increase batch size

Answer

Increase learning rate

Answer

Gradient clipping

Question 11

A data scientist is training a multi-class classifier with 10 classes. The training log shows the above output for the first two epochs. What is the most likely cause?

Accepted Answer

The learning rate is set to zero. When the learning rate is set to zero, the optimizer makes no updates to the model weights regardless of the computed gradients. The training loss remains constant across epochs because the parameters never change, which matches the log showing identical loss values for both epochs. This is a common debugging scenario where a misconfigured learning rate prevents any learning from occurring.

Answer

Batch normalization is disabled

Answer

The dataset is imbalanced

Answer

The model is overfitting

Question 12

A team is reviewing a neural network model summary. The input layer expects 784 features (e.g., 28x28 images). How many parameters does the first dense layer have?

Accepted Answer

100,352. The first dense layer has 784 input features and 128 output units (a common default). Each of the 784 inputs connects to each of the 128 neurons, giving 784 * 128 = 100,352 weight parameters, plus 128 bias parameters (one per neuron), for a total of 100,480 parameters. However, the question asks for the number of parameters in the dense layer itself, and the correct answer is 100,352, which corresponds to the weight parameters only, as biases are often listed separately or the layer uses no bias. In typical Keras summaries, the parameter count for a Dense layer with bias is (input_dim * units) + units, but here the provided correct answer matches the weight count alone, indicating the model summary excludes biases or uses a bias-less configuration.

Answer

100,224

Answer

109,258

Answer

8,256

Question 13

A data scientist is training a neural network to classify images of handwritten digits. The model achieves 99% accuracy on training data but only 85% on validation data. Which technique should the scientist apply first to address this issue?

Accepted Answer

Apply L2 regularization to the network weights. The model shows high training accuracy (99%) but lower validation accuracy (85%), which is a classic sign of overfitting. L2 regularization (option C) adds a penalty term to the loss function proportional to the squared magnitude of the weights, discouraging the network from learning overly complex patterns that do not generalize. This directly addresses overfitting without reducing the model's capacity too aggressively.

Answer

Remove one or more hidden layers from the network

Answer

Increase the number of training epochs

Answer

Add more features to the input data

Question 14

A company is deploying a machine learning model to predict customer churn. The dataset is highly imbalanced (95% non-churn, 5% churn). The model achieves 96% accuracy, but the F1-score for the churn class is only 0.2. Which metric should the team prioritize to evaluate model performance for this business problem?

Accepted Answer

F1-score. In a highly imbalanced dataset (95% non-churn, 5% churn), accuracy is misleading because a model can achieve 96% accuracy by simply predicting the majority class for all instances. The F1-score, which is the harmonic mean of precision and recall, specifically measures the model's performance on the minority (churn) class. A low F1-score of 0.2 indicates the model fails to correctly identify churners, which is the critical business outcome, making F1-score the correct metric to prioritize.

Answer

Accuracy

Answer

Log loss

Answer

AUC-ROC

Question 15

An autonomous vehicle system uses a deep reinforcement learning agent to navigate. The agent's reward function gives +1 for reaching the destination and -0.1 for each time step. After training, the agent learns to circle the block repeatedly without reaching the destination. Which modification is most likely to fix this behavior?

Accepted Answer

Increase the time penalty to -1 per step. The agent learns to circle the block because the cumulative penalty for each time step (-0.1) is too small relative to the reward for reaching the destination (+1). By increasing the time penalty to -1 per step, the agent will incur a much larger cost for delaying, making it optimal to reach the destination quickly rather than looping indefinitely. This directly addresses the reward structure imbalance that causes the undesirable behavior.

Answer

Increase the reward for reaching the destination to +10

Answer

Use a discount factor closer to 0

Answer

Add a penalty for each turn the vehicle makes

Question 16

A machine learning engineer is building a spam filter. The dataset contains 10,000 emails, of which 1,000 are spam. The engineer decides to use a Random Forest classifier. Which preprocessing step is most critical to ensure the model generalizes well to new, unseen emails?

Accepted Answer

Split the data into training and testing sets before any other preprocessing. Option C is correct because splitting the data into training and testing sets before any other preprocessing prevents data leakage. If preprocessing like normalization or PCA is applied to the entire dataset first, the test set information influences the training process, leading to overly optimistic performance estimates and poor generalization to new, unseen emails.

Answer

Apply Principal Component Analysis (PCA) to reduce dimensionality

Answer

Normalize the numerical features to have zero mean and unit variance

Answer

Encode all features using one-hot encoding

Question 17

Which TWO techniques are commonly used to prevent overfitting in deep neural networks?

Accepted Answer

Dropout. Dropout is a regularization technique that randomly drops a fraction of neurons during training, which prevents the network from relying too heavily on any single neuron and forces it to learn more robust features. This reduces overfitting by introducing noise and effectively training an ensemble of sub-networks.

Answer

Using a larger learning rate

Answer

L1 regularization

Answer

Increasing the number of layers

Question 18

Refer to the exhibit. A data scientist is training a binary classifier. Based on the training log, which problem is the model experiencing?

Accepted Answer

Overfitting. The training log shows that the model's training accuracy continues to improve while the validation accuracy plateaus or degrades after a certain number of epochs. This divergence between training and validation performance is the hallmark of overfitting, where the model memorizes the training data noise rather than learning generalizable patterns.

Answer

Underfitting

Answer

Data leakage

Answer

Vanishing gradient

Question 19

A healthcare startup is developing a deep learning model to detect diabetic retinopathy from retinal fundus images. The dataset contains 50,000 images, but only 5% are labeled as positive for the disease. The team uses a convolutional neural network (CNN) with a final sigmoid layer and binary cross-entropy loss. After training for 20 epochs, the model achieves 95% accuracy on the test set, but the recall for the positive class is only 10%. The team suspects the model is biased toward the negative class due to class imbalance. The data is stored in a secure environment, and no additional labeled data can be obtained. The team has access to the following techniques: oversampling the minority class, undersampling the majority class, using class weights in the loss function, applying data augmentation, and using a different architecture. Which course of action is most likely to improve recall for the positive class while maintaining reasonable overall performance?

Accepted Answer

Assign higher class weights to the positive class in the loss function. Assigning higher class weights to the positive class in the loss function directly penalizes misclassifications of the minority class during training. This forces the model to pay more attention to positive samples without altering the dataset distribution, which is critical when no additional labeled data can be obtained and the data is in a secure environment. It improves recall by increasing the gradient contribution from positive samples, while maintaining overall performance because the model still sees the original data distribution.

Answer

Undersample the majority class to balance the dataset

Answer

Oversample the minority class using synthetic image generation

Answer

Replace the CNN with a transformer-based architecture

Question 20

A retail company uses a gradient boosting model to predict customer lifetime value (CLV). The model currently uses 50 features including purchase history, demographics, and web behavior. The model's RMSE on the test set is 120. The data science team wants to improve the model's accuracy without increasing training time significantly. They have access to additional data: customer support interaction logs (text), social media sentiment (text), and third-party credit scores (numeric). They also have the ability to perform feature engineering, hyperparameter tuning, and ensemble methods. Which approach is most likely to yield the best improvement in predictive performance with minimal increase in training time?

Accepted Answer

Engineer new features such as average purchase value and recency. Option D is correct because engineering domain-relevant features like average purchase value and recency directly captures the underlying behavioral patterns that drive customer lifetime value, often providing a higher signal-to-noise ratio than adding raw text or third-party data. This approach leverages existing data without significantly increasing the feature dimensionality or training time, unlike adding TF-IDF vectors which would dramatically expand the feature space and slow training.

Answer

Add the customer support text as a feature using TF-IDF vectors

Answer

Use an ensemble of gradient boosting and random forest models

Answer

Perform hyperparameter tuning using grid search