Question

A finance team needs to automatically reject expense reports that exceed a fixed dollar limit and that are submitted outside company-defined categories. The rules are explicit, rarely change, and every decision must be fully explainable to auditors. A developer suggests building this with a deep learning model. What is the most appropriate recommendation?

Accepted Answer

Use a deterministic rules-based system instead, since the logic is well-defined and does not require learning patterns from data

Answer

Train a supervised deep learning model on historical expense reports to predict approvals

Answer

Use an unsupervised clustering model to group expense reports before applying limits

Answer

Use reinforcement learning so the system improves its rejection decisions over time