Financial Fraud Detection Approach Based on Firefly Optimization Algorithm and Support Vector Machine

Figure 5

Comparison of error rates between the proposed algorithm and three other search algorithms (BSVM, GSVM, and GASVM). The error rates of the BSVM and GSVM algorithms are about to 14.50%. However, in the case of the GASVM algorithm, which has a high error rate (15.07%), it means the performance of a fraud detection algorithm can be decreased. The proposed algorithm has achieved a much lower error rate (14.34%). It means that the financial fraud detection correctness is high, and the most accurate way to detect fraud is by FFSVM.