Question

A financial services SOC has a large historical dataset of transactions, each already labeled by fraud analysts as either 'fraudulent' or 'legitimate.' Leadership wants an AI-powered analytic capability that can automatically assign one of these two categories to each new incoming transaction in real time. Which machine learning technique should you recommend to meet this specific need?

Accepted Answer

Supervised classification using the labeled historical transactions as training data

Answer

Unsupervised clustering to group transactions by behavioral similarity

Answer

Reinforcement learning to reward the agent for exploring transaction states

Answer

Dimensionality reduction to compress transaction features before storage