Web services are utilized in nearly every aspect of modern life. The maintenance and testing costs of these services are quite high due to difficulties such as automatic service discovery, ultra-late binding, and cost per usage. One way to reduce these costs is to generate test scenarios automatically. However, a large number of requests are generated automatically when testing the application. This causes inefficient resource utilization, particularly due to challenges such as high usage costs. Therefore, it is possible to verify if these randomly generated test scenarios are valid. In this study, the hyperparameters of the deep neural network, in which the inputs of 8 different services are classified as “valid“ and “faulty,“ were optimized with the artificial bee colony algorithm. According to the obtained results, thanks to the hyperparameter optimization carried out using the artificial bee colony algorithm, improvements were achieved in both cross-validation and test values.