Incremental learning strategies for credit cards fraud detection

Every second, thousands of credit or debit card transactions are processed in financial institutions. This extensive amount data and its sequential nature make the problem fraud detection particularly challenging. Most analytical strategies used production still based on batch learning, which is inadequate for two reasons: Models quickly become outdated require sensitive storage. The evolving bank enshrines importance having up-to-date models, retention makes companies vulnerable to infringements European General Data Protection Regulation. For these reasons, evaluating incremental learning recommended. paper designs evaluates solutions real-world systems. aim demonstrate competitiveness over conventional approaches and, consequently, improve accuracy employing ensemble diversity transfer learning. An experimental analysis conducted a full-scale case study including five months e-commerce made available by our industry partner, Worldline.