Different AI Models and Development Processes AIE Practice Questions (Page 3)

Questions 11–15

1. 11application · medium

   A bank wants to build a model to approve or reject loan applications. They have a small dataset of 2,000 past applications with labels. The model must be highly accurate and avoid false approvals. Which model selection approach is most appropriate?

   • AUse a simple logistic regression model because it is interpretable and works well with small datasets
   • BUse reinforcement learning to learn the approval policy from historical decisions
   • CUse an unsupervised clustering model to group similar loans
   • DUse a deep neural network because it always achieves the highest accuracy
   Reveal Solution Discussion Select an answer first
2. 12application · medium

   A retail company wants to build a model that segments its customers into distinct groups based on purchasing behavior so that marketing campaigns can be tailored to each group. The data science team has a large dataset of historical transactions with no predefined labels for customer segments. Which model type should they use?

   • AReinforcement learning model
   • BUnsupervised clustering model
   • CSupervised classification model
   • DSemi-supervised regression model
   Reveal Solution Discussion Select an answer first
4. 13expert · hard

   A financial firm wants to build a model to detect insider trading. They have a small labeled dataset of confirmed cases and a large amount of unlabeled trading data. The cost of missing a case is extremely high, but false accusations can ruin careers. Which approach best balances these constraints?

   • ASupervised learning with high recall threshold
   • BReinforcement learning to optimize trading decisions
   • CUnsupervised anomaly detection on all trading data
   • DSemi-supervised learning using labeled and unlabeled data, with a precision-focused evaluation
   Reveal Solution Discussion Select an answer first
5. 14expert · hard

   A self-driving car company is training a model to decide when to change lanes. The model receives rewards for safe and efficient lane changes and penalties for risky maneuvers. The team is debating between using a model-free reinforcement learning algorithm and a supervised learning approach. What is the key reason to prefer reinforcement learning in this scenario?

   • ASupervised learning cannot be evaluated
   • BReinforcement learning does not require any data
   • CSupervised learning is not capable of processing sensor data
   • DReinforcement learning can learn from delayed rewards and interactions with the environment
   Reveal Solution Discussion Select an answer first
6. 15foundation · easy

   A data scientist has a dataset of labeled customer emails, each marked as 'spam' or 'not spam'. They want to train a model to classify new emails. Which type of AI model is most appropriate for this task?

   • ASupervised learning model
   • BUnsupervised learning model
   • CSemi-supervised learning model
   • DReinforcement learning model
   Reveal Solution Discussion Select an answer first