AI0-001 AI Concepts and Techniques Practice Test 5 — 15 Questions

Question 1

A team is deploying a sentiment analysis model that must achieve high precision and high recall. They have a labeled dataset of 10,000 samples. They want to minimize overfitting. Which THREE actions are most appropriate? (Select THREE.)

Choose an answer to begin — your selection is scored in the full session.

Accepted Answer

Apply L2 regularization to the model weights. Option B is correct because applying L2 regularization adds a penalty proportional to the squared magnitude of the model weights to the loss function, which constrains weight growth and directly reduces overfitting, helping the model generalize and maintain both precision and recall on unseen data. Option C is correct because dropout layers randomly deactivate a fraction of neurons during each training iteration, preventing the network from relying on specific co-adapted units and acting as an effective regularizer that lowers overfitting. Option E is correct because augmenting the 10,000-sample training set with synthetic examples increases data diversity and effective sample size, which is a standard technique to improve generalization and reduce overfitting when labeled data is limited. Option A is not appropriate because decreasing the learning rate only affects optimization step size and convergence stability, not the model's capacity to overfit, and can even slow convergence without regularizing. Option D is not appropriate because increasing the training batch size changes gradient estimation variance and training dynamics but does not by itself prevent overfitting; in fact, very large batches can sometimes harm generalization.

Answer

Decrease the learning rate

Answer

Increase the training batch size

Question 2

A company wants to classify images of products into categories. They have a large dataset of labeled images. Which TWO types of neural networks are most suitable for this task? (Select TWO.)

Accepted Answer

Convolutional Neural Network (CNN). Option B, Convolutional Neural Network (CNN), is correct because CNNs use convolutional and pooling layers to exploit spatial locality and translation invariance in images, making them the classic architecture for supervised image classification with large labeled datasets. Option D, Transformer (e.g., Vision Transformer), is also correct because a Vision Transformer splits an image into patches, embeds them, and applies self-attention to model global relationships, achieving state-of-the-art results on image classification when sufficient labeled data (or pretraining) is available. Option A, Generative Adversarial Network (GAN), is not appropriate here because GANs are generative models that learn to synthesize data via a generator-discriminator game rather than directly performing supervised category classification. Option C, Recurrent Neural Network (RNN), is unsuitable because RNNs are designed for sequential data such as text or time series and do not natively capture the 2D spatial structure of images. Option E, Multi-layer Perceptron (MLP), is not the best choice because fully connected layers ignore spatial hierarchy and typically underperform CNNs or Transformers on large-scale image classification.

Answer

Generative Adversarial Network (GAN)

Answer

Recurrent Neural Network (RNN)

Answer

Multi-layer Perceptron (MLP)

Question 3

A data scientist is building a model to predict whether a transaction is fraudulent. The dataset has 99.9% legitimate transactions and 0.1% fraudulent ones. Which evaluation metric is MOST appropriate to assess model performance given this class imbalance?

Accepted Answer

F1-score. With 99.9% legitimate transactions and only 0.1% fraudulent ones, accuracy would be misleadingly high (99.9%) even if the model never predicts fraud. The F1-score is the harmonic mean of precision and recall, making it robust to class imbalance by penalizing both false positives and false negatives. This makes it the most appropriate metric for evaluating fraud detection performance.

Answer

BLEU score

Answer

Accuracy

Answer

Perplexity

Question 4

Which machine learning paradigm involves training an agent to make decisions by interacting with an environment and receiving rewards or penalties based on its actions?

Accepted Answer

Reinforcement learning. Reinforcement learning (RL) is the correct paradigm because it explicitly involves an agent learning a policy through trial-and-error interactions with an environment, receiving scalar reward signals (positive or negative) to maximize cumulative reward. This matches the question's description of making decisions based on rewards or penalties, which is the defining characteristic of RL, as opposed to learning from labeled data or discovering hidden patterns without feedback.

Answer

Unsupervised learning

Answer

Supervised learning

Answer

Self-supervised learning

Question 5

A developer is fine-tuning a large language model for a legal document summarization task. They notice that during training, the loss decreases rapidly in the first few epochs but then plateaus with high variance. Which hyperparameter adjustment is MOST likely to help stabilize training?

Accepted Answer

Decrease the learning rate. A high-variance loss plateau after rapid initial convergence typically indicates that the learning rate is too large, causing the optimizer to overshoot the minima and oscillate. Decreasing the learning rate allows smaller, more stable weight updates, reducing variance and enabling smoother convergence.

Answer

Add L1 regularization

Answer

Increase the batch size

Answer

Increase the number of epochs

Question 6

A team is deploying a sentiment analysis model that must achieve high precision and high recall. They have a labeled dataset of 10,000 samples. They want to minimize overfitting. Which THREE actions are most appropriate? (Select THREE.)

Choose an answer to begin — your selection is scored in the full session.

Accepted Answer

Fine-tune the existing model on a corpus of code-switched text. Fine-tuning the existing model on a corpus of code-switched text adapts the model to the new language pattern efficiently.

Answer

Use a larger base model without additional training

Answer

Apply data augmentation by translating all code-switched posts to English

Answer

Train a new model from scratch on a mix of English and code-switched data

Question 7

Which neural network architecture is specifically designed to process sequential data, such as time series or sentences, by maintaining a hidden state that captures information about previous inputs?

Accepted Answer

Recurrent Neural Network (RNN). Recurrent Neural Networks (RNNs) are specifically designed for sequential data because they maintain a hidden state that is updated at each time step, allowing information about previous inputs to persist and influence current and future outputs. This feedback loop makes them ideal for tasks like time series forecasting, natural language processing, and speech recognition, where order and context matter.

Answer

Transformer

Answer

Convolutional Neural Network (CNN)

Answer

Multi-layer Perceptron (MLP)

Question 8

A company wants to automatically group customer support tickets into categories (e.g., billing, technical, account) without pre-labeled data. Which machine learning approach should they use?

Accepted Answer

Unsupervised clustering using K-means. The company has no pre-labeled data, which means supervised learning (which requires labeled examples) is not feasible. Unsupervised clustering, such as K-means, groups data points into clusters based on feature similarity without needing any labels, making it ideal for automatically discovering categories like billing, technical, or account from raw ticket text.

Answer

Supervised classification with logistic regression

Answer

Semi-supervised learning with a small labeled set

Answer

Reinforcement learning with a reward function

Question 9

A generative AI model produces images from text prompts. The outputs are often blurry and lack fine details. Which model type is MOST likely being used, and which improvement would best address this issue?

Accepted Answer

Variational Autoencoder (VAE); switch to a diffusion model. Variational Autoencoders (VAEs) are known for producing blurry outputs because their loss function (ELBO) encourages pixel-wise averaging, which smooths out fine details. Diffusion models, by contrast, iteratively denoise a random field, learning to reconstruct high-frequency details through a multi-step reverse process, directly addressing the blurriness issue.

Answer

Variational Autoencoder (VAE); switch to a Generative Adversarial Network (GAN)

Answer

Generative Adversarial Network (GAN); increase the discriminator's capacity

Answer

Diffusion model; use a larger batch size during training

Question 10

Which of the following best describes the difference between narrow AI and general AI?

Accepted Answer

Narrow AI is designed for a specific task; general AI aims to perform any cognitive task a human can.. Narrow AI specializes in one task, while general AI would possess human-like cognitive abilities across domains.

Answer

Narrow AI relies on supervised learning; general AI uses unsupervised learning exclusively.

Answer

Narrow AI requires large datasets; general AI can learn from few examples.

Answer

Narrow AI can perform any intellectual task; general AI is limited to specific tasks.

Question 11

A team is training a deep learning model for image classification. They observe that training accuracy is high but validation accuracy is low, indicating overfitting. Which TWO techniques should they apply to reduce overfitting? (Select TWO)

Accepted Answer

Use L2 regularization. Option A, using L2 regularization, is correct because adding a weight-decay penalty (e.g., lambda * sum of squared weights) to the loss function constrains the magnitude of the model's weights, discouraging the network from fitting noise in the training set and thereby reducing overfitting. Option C, adding dropout layers, is correct because dropout randomly deactivates a fraction of neurons (e.g., p=0.5) during each training step, preventing units from co-adapting and forcing the network to learn more robust, generalizable features. Option B, increasing the learning rate, is not appropriate because a larger learning rate typically causes unstable or divergent training rather than reducing overfitting. Option D, increasing the number of layers, would raise model capacity and generally worsen overfitting. Option E, reducing training data size, would make overfitting more severe by giving the model even less data to generalize from.

Answer

Increase learning rate

Answer

Increase the number of layers

Answer

Reduce training data size

Question 12

An AI engineer is fine-tuning a transformer-based language model for a domain-specific task. They want to improve the model's factual accuracy and reduce hallucinations. Which THREE strategies should they consider? (Select THREE)

Accepted Answer

Fine-tune the model on a curated domain-specific corpus. Option B is correct because fine-tuning on a curated, high-quality domain-specific corpus injects accurate, task-relevant knowledge into the model's weights, which directly improves factual grounding and reduces the likelihood of fabricating domain facts. Option D is correct because chain-of-thought prompting encourages the model to decompose complex queries into intermediate reasoning steps, which empirically reduces errors and unsupported claims on multi-step or reasoning-heavy tasks. Option E is correct because Retrieval-Augmented Generation grounds generation in externally retrieved, up-to-date documents at inference time, so the model conditions its output on verifiable evidence rather than relying solely on parametric memory, which is the standard technique for reducing hallucinations. Option A is not appropriate because simply increasing the context window beyond the training limit is not a supported or effective strategy—models cannot attend beyond their trained positional range without architectural changes or extrapolation techniques, and a larger window alone does not improve factual accuracy. Option C is incorrect because raising the temperature increases sampling randomness and diversity, which typically increases hallucination rather than reducing it; lower temperature is preferred for factual accuracy.

Answer

Increase the model's context window size beyond the training limit

Answer

Use a higher temperature setting during generation

Question 13

A data scientist needs to select a regression model to predict house prices. The dataset contains many features, some of which are irrelevant. Which TWO algorithms are BEST suited for this scenario, and why? (Select TWO)

Accepted Answer

Lasso regression (L1 regularization). Lasso regression (L1 regularization) is correct because its L1 penalty drives the coefficients of irrelevant features exactly to zero, performing automatic feature selection and yielding a sparse, interpretable model well suited to a dataset with many useless predictors. Random Forest is correct because its ensemble of decorrelated decision trees handles high-dimensional feature spaces robustly, captures non-linear relationships and interactions between features, and provides built-in feature-importance scores that tolerate irrelevant variables without overfitting as easily as a single model. Linear regression is not appropriate because it uses all features with no regularization, so irrelevant predictors inflate variance and degrade generalization. Ridge regression shrinks coefficients toward zero but never eliminates them, so it does not perform feature selection. K-Nearest Neighbors is distance-based and suffers from the curse of dimensionality, making it a poor choice when many features are irrelevant.

Answer

Ridge regression (L2 regularization)

Answer

Linear regression

Answer

K-Nearest Neighbors

Question 14

A company is deploying an LLM-powered application that answers questions based on internal documents. They want to minimize prompt injection attacks where users trick the model into ignoring instructions. Which THREE measures should they implement? (Select THREE)

Accepted Answer

Use a system-level prompt that clearly defines allowed behavior and boundaries. Option A is correct because a system-level prompt that clearly defines allowed behavior and boundaries establishes the model's operating rules and helps it distinguish trusted instructions from untrusted user content, which is a core defense against prompt injection. Option D is correct because a separate classifier trained to detect and block injection attempts adds an independent detection layer that can catch malicious prompts before they reach the LLM, reducing the chance that a user overrides system instructions. Option E is correct because sanitizing user inputs to remove special tokens or known injection patterns prevents attackers from injecting delimiters, role markers, or instruction-like text that could confuse the model's instruction hierarchy. Option B is not correct because setting temperature to 0.0 only makes output more deterministic and does not prevent prompt injection. Option C is not correct because allowing the model to execute arbitrary code from user prompts dramatically increases risk and is the opposite of a security control.

Answer

Set temperature to 0.0 for all queries

Answer

Allow the model to execute any code from user prompts for flexibility

Question 15

A company wants to use machine learning to recommend products to customers based on their purchase history. Which TWO techniques are appropriate for this task? (Select TWO)

Accepted Answer

Collaborative filtering. Collaborative filtering (A) is correct because it is the canonical recommendation technique that leverages patterns across users' purchase histories (user-item interaction matrices) to predict products a customer is likely to buy, either via user-based or item-based similarity. K-Nearest Neighbors (C) is also correct because it can be applied to recommendation by finding the k most similar users or items based on historical purchase vectors and aggregating their preferences to generate recommendations. PCA (B) is a dimensionality-reduction technique, not a recommender, though it may be used as preprocessing. Naive Bayes (D) is a probabilistic classifier for labeled categories such as spam detection, not suited to ranking product recommendations from purchase history. Linear regression (E) predicts a continuous numeric value and does not model user-item preference relationships needed for product recommendation.

Answer

Principal Component Analysis (PCA)

Answer

Naive Bayes

Answer

Linear regression