Fraud Detection in Payment System using Genetic Algorithm

The increasing volume and speed of digital payments have amplified the challenge of fraud detection. Traditional rule-based systems are often rigid and unable to keep pace with evolving fraudulent tactics. This paper presents a novel approach to fraud detection in payment systems using a genetic algorithm (GA) to address key challenges such as imbalanced datasets and the need for adaptable models. By mimicking the process of natural selection, the GA iteratively refines a population of potential solutions (e.g., feature subsets or classifier weights). Each solution's "fitness" is evaluated based on a custom objective function that may prioritize high detection accuracy while minimizing false positives, an important trade-off in financial fraud detection. Through selection, crossover, and mutation, the algorithm converges on an optimized model that can effectively distinguish between legitimate and fraudulent transactions.