Anomaly detection in financial and business data
2025 CMU Accounting Mini Conference on
Accounting Applications of Information Science and Economics
Christos Faloutsos, CMU CSD
Aug. 16, 2025
Abstract
Given information about who pays whom and when, how can we find suspicious entities? We give several case studies where computer science and visualization tools helped spot fraudulent accounts, 'smurfing' accounts for potential money laundering, and accounts with weird temporal behavior. We demonstrate these tools on accounting and e-commerce data, and we also emphasize the importance of explanation and visualization, like our 'CallMine' tool. We avoid mathematical proofs and instead focus on the intuition behind these tools.
Slides
Last updated: Aug. 14, 2025, by Christos Faloutsos