AI0-001 Practice Test 17 — 15 Questions

Question 1

A data science team is preparing a dataset for a binary classification model to detect fraudulent transactions. The dataset has 99% legitimate and 1% fraudulent examples. Which TWO techniques should the team apply to improve model performance on the minority class?

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

Accepted Answer

Use class weights in the loss function. Option A is correct because assigning class weights in the loss function (e.g., class_weight='balanced' in scikit-learn or a weighted binary cross-entropy) penalizes misclassification of the 1% fraudulent class more heavily, directly countering the 99:1 imbalance during training. Option B is correct because SMOTE (Synthetic Minority Over-sampling Technique) generates synthetic minority-class samples by interpolating between existing fraudulent examples and their k-nearest neighbors, increasing minority representation and helping the model learn the fraudulent decision boundary. Option C is not selected because random undersampling discards potentially useful majority-class data and can increase variance, making it a less preferred technique here. Option D is not selected because z-score normalization only rescales feature distributions and does nothing to address class imbalance. Option E is not selected because shuffling prevents ordering bias and train/test leakage but has no effect on the skewed class ratio.

Answer

Undersample the majority class randomly

Answer

Apply data normalisation (z-score) to all features

Answer

Randomly shuffle the dataset to prevent train/test leakage

Question 2

A recommendation system for an e-commerce platform is experiencing a high false positive rate in its anomaly detection module, causing legitimate transactions to be flagged as fraudulent. The team wants to reduce false positives without significantly increasing false negatives. Which action is MOST effective?

Accepted Answer

Increase the anomaly detection threshold. Increasing the anomaly detection threshold makes the model more conservative about flagging transactions as anomalous, which directly reduces false positives (legitimate transactions incorrectly flagged). Because the threshold is raised, only more extreme deviations trigger an alert, so some true anomalies may be missed, but the question explicitly accepts a small increase in false negatives as a tradeoff.

Answer

Decrease the anomaly detection threshold

Answer

Use a different anomaly detection algorithm

Answer

Increase the size of the training dataset

Question 3

A team is building a regression model to predict house prices. Which data transformation is most appropriate if the target variable exhibits right skewness?

Accepted Answer

Log transformation. Log transformation is the most appropriate technique for right-skewed target variables because it compresses the long tail, making the distribution more symmetric and closer to Gaussian. This stabilizes variance and often improves the performance of regression models that assume normally distributed errors, such as linear regression.

Answer

Principal component analysis (PCA)

Answer

Standardization (Z-score)

Answer

One-hot encoding

Question 4

A machine learning engineer is preparing to train a deep neural network for image classification. To avoid overfitting, which TWO techniques should the engineer apply? (Select TWO.)

Accepted Answer

Use dropout regularization.. Dropout regularization is a technique that randomly drops a fraction of neurons during training, which prevents the network from relying too heavily on any single neuron and reduces co-adaptation. This acts as a form of ensemble learning and significantly reduces overfitting by improving generalization.

Answer

Increase the number of layers.

Answer

Remove all non-linear activation functions.

Answer

Reduce the training dataset size.

Question 5

An AI engineer is designing a system to detect unusual patterns in network traffic that may indicate a security breach. The system should learn from normal traffic patterns and flag deviations. Which machine learning approach is MOST appropriate?

Accepted Answer

Unsupervised anomaly detection. Unsupervised anomaly detection is the most appropriate approach because the system must learn 'normal' traffic patterns from unlabeled data and then flag deviations without requiring pre-labeled examples of attacks. This aligns with the core requirement of detecting unknown or novel security breaches, which supervised methods cannot handle due to the lack of labeled attack data.

Answer

Reinforcement learning with reward shaping

Answer

Supervised classification using logistic regression

Answer

Semi-supervised learning with a small labeled set

Question 6

Which TWO of the following are common techniques to improve the transparency and interpretability of an AI model?

Accepted Answer

Generate SHAP (SHapley Additive exPlanations) values. Option A is correct because SHAP (SHapley Additive exPlanations) values, grounded in cooperative game theory, assign each feature a quantitative contribution to a prediction, providing both global and local interpretability for any model. Option E is correct because LIME (Local Interpretable Model-agnostic Explanations) approximates a complex model's behavior around a single prediction with a simple, interpretable surrogate model, exposing which features drove that decision. Both techniques are model-agnostic post-hoc explanation methods specifically designed to make AI outputs transparent and interpretable. Option B is not a transparency technique: differential privacy adds calibrated noise to protect individual records, trading accuracy for privacy rather than explaining model behavior. Option C is not inherently an interpretability technique; random forests are ensembles whose many trees are typically less transparent than a single decision tree. Option D is incorrect because increasing complexity with deep neural networks generally reduces interpretability rather than improving it.

Answer

Use differential privacy to add noise to training data

Answer

Implement a random forest algorithm

Answer

Use deep neural networks to increase model complexity

Question 7

A machine learning engineer is training a neural network for image classification. The training loss decreases slowly and the model accuracy improves only marginally each epoch. Which hyperparameter adjustment is MOST likely to accelerate convergence?

Accepted Answer

Increase the learning rate. The training loss decreasing slowly and accuracy improving marginally each epoch indicates that the learning rate is too small, causing the optimizer to take very small steps toward the minimum of the loss function. Increasing the learning rate allows the optimizer to take larger steps per update, which accelerates convergence. Option C is correct because adjusting the learning rate directly addresses the step size in gradient descent.

Answer

Add more hidden layers

Answer

Increase the batch size

Answer

Decrease the number of epochs

Question 8

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 9

A healthcare startup deploys an AI model to predict patient readmission rates. An internal audit reveals that the model consistently underestimates readmission risk for non-native English speakers. According to AI ethics principles, what is the most appropriate course of action?

Accepted Answer

Retrain the model with a more representative dataset that includes diverse language backgrounds. It directly addresses the root cause of the bias: the training data lacks sufficient representation from non-native English speakers, leading to systematic underestimation of readmission risk for that group. Retraining with a more representative dataset aligns with the AI ethics principle of fairness by ensuring the model learns patterns across all demographic groups equally, rather than masking the issue with disclaimers or manipulating sample sizes.

Answer

Add a confidence score disclaimer to model outputs

Answer

Reduce the sample size of non-native English speakers to balance the dataset

Answer

Continue using the model as is, since overall accuracy is acceptable

Question 10

A company is training a model on proprietary data and wants to prevent data poisoning. Which TWO practices are most important? (Select TWO.)

Accepted Answer

Implementing access controls on the training dataset. Option A (Implementing access controls on the training dataset) is correct because data poisoning requires an adversary to inject or modify training samples, and strict authentication/authorization (e.g., IAM roles, least-privilege permissions on the S3 bucket or data lake) prevents unauthorized parties from tampering with the proprietary dataset in the first place. Option B (Validating the integrity of training data) is correct because even with access controls, data can be corrupted or subtly altered, so integrity checks such as cryptographic hashes/checksums, provenance tracking, and outlier or anomaly detection help detect poisoned or tampered samples before they influence the model. Option C (Using a larger model) does not belong because model capacity has no bearing on whether poisoned data enters the pipeline and can even make a model more susceptible to memorizing malicious samples. Option D (Increasing training epochs) does not belong because more training iterations only reinforce whatever data is present, potentially amplifying the effect of poisoned samples rather than preventing them. Option E (Using homomorphic encryption) does not belong because it protects data confidentiality during computation, not the authenticity or integrity of the training data against poisoning.

Answer

Using a larger model

Answer

Increasing training epochs

Answer

Using homomorphic encryption

Question 11

A data scientist suspects a model extraction attack on their deployed classifier. Which TWO indicators are MOST consistent with such an attack? (Select two.)

Accepted Answer

A large number of queries from a single IP address over a short period. Option C is correct because model extraction (model stealing) attacks require the adversary to submit a very high volume of prediction requests to the deployed classifier, often from a single source IP, in order to gather enough input-output pairs to train a substitute model; a burst of queries from one IP in a short window is a classic volumetric indicator. Option E is correct because effective extraction depends on sampling the model's decision space broadly, so queries that systematically span a wide, diverse range of inputs (rather than a narrow slice) are consistent with an attacker trying to approximate the full decision boundary. Option A is not correct because SQL injection targets database-backed application inputs and is unrelated to stealing a model's parameters or behavior. Option B is not correct because repeatedly requesting the same prediction yields redundant, low-information samples and is more indicative of caching, retry, or probing behavior than systematic extraction. Option D is not correct because attempts to reveal system prompts are prompt-injection or prompt-leaking attacks against LLM applications, not model extraction of a classifier.

Answer

Queries that include SQL injection attempts

Answer

Queries that repeatedly ask for the same prediction

Answer

Queries with special characters attempting to reveal system prompts

Question 12

An organization is adopting a third-party pre-trained language model for internal use. To assess supply chain security, which document should they request to understand the components and dependencies of the model?

Accepted Answer

Software Bill of Materials (SBOM). A Software Bill of Materials (SBOM) is the correct document for assessing supply chain security because it provides a detailed, machine-readable inventory of all components, libraries, and dependencies used to build the model. This allows the organization to identify known vulnerabilities, licensing risks, and transitive dependencies, which is essential for evaluating the security posture of a third-party pre-trained model.

Answer

OWASP LLM Top 10

Answer

Model card

Answer

Data flow diagram

Question 13

An organization wants to assess the security of its custom LLM application before production release. Which practice involves simulating attacks to identify vulnerabilities?

Accepted Answer

Red teaming. Red teaming (Option D) is the correct practice for simulating attacks to identify vulnerabilities in a custom LLM application. This involves ethical hackers or security experts actively probing the system with adversarial inputs, such as prompt injection, jailbreaking, or data poisoning attempts, to uncover weaknesses before production release. It directly tests the application's resilience against real-world attack vectors, aligning with the AI Security domain's focus on proactive threat assessment.

Answer

Blue teaming

Answer

Model validation

Answer

Data sanitization

Question 14

A data scientist is training a customer churn prediction model using sensitive customer data. To comply with data privacy regulations, they want to minimize the risk of membership inference attacks. Which TWO techniques should they consider?

Accepted Answer

Use techniques to reduce overfitting, such as regularization or simpler models. Option C is correct because membership inference attacks exploit a model's tendency to overfit to its training data: an overfit model behaves very differently on training versus non-training samples, making it easier for an attacker to determine whether a specific record was in the training set. Reducing overfitting via L1/L2 regularization, dropout, early stopping, or simpler model architectures shrinks this generalization gap and thus lowers membership leakage. Option D is correct because differential privacy during training (e.g., DP-SGD with gradient clipping and calibrated noise) provides a formal, quantifiable guarantee that any single individual's presence or absence in the training set has only a bounded effect on the model's output, which directly limits what a membership inference attack can infer. Option A is not correct because k-fold cross-validation is a model-evaluation and hyperparameter-tuning technique; it does not by itself reduce the information a released model leaks about its training records. Option B is not correct because hiding confidence scores only removes one attack signal while the model still exposes predictions and gradients/behaviors that can be exploited; it is not a principled privacy defense. Option E is not correct because adding augmented or synthetic data does not provide any privacy guarantee and can even increase memorization of the original sensitive records.

Answer

Use k-fold cross-validation to improve model accuracy

Answer

Deploy the model as a black-box API with no confidence scores

Answer

Increase training data size through data augmentation

Question 15

A financial institution uses an AI model to approve loan applications. The model was trained on historical data that included biased lending practices. The bank's ethics committee wants to mitigate bias without removing protected attributes. Which approach best balances fairness and model performance?

Accepted Answer

Apply adversarial debiasing during training. Adversarial debiasing is the best approach because it directly optimizes the model to reduce bias during training while preserving predictive accuracy. It uses an adversarial network that tries to predict the protected attribute from the model's predictions, forcing the main model to learn representations that are less correlated with that attribute. This allows the bank to keep protected attributes in the data (as required by the ethics committee) while actively mitigating bias.

Answer

Retrain the model using a balanced dataset

Answer

Remove all protected attributes from the training data

Answer

Post-process model outputs to adjust for demographic parity