Akademik Veri Yönetim Sistemi
20th International Conference on Future Networks and Communications, FNC 2025 / 22nd International Conference on Mobile Systems and Pervasive Computing, MobiSPC 2025 / 15th International Conference on Sustainable Energy Information Technology, SEIT 2025, Leuven, Belçika, 4 - 06 Ağustos 2025, cilt.265, ss.566-571, (Tam Metin Bildiri)
Periodic updates and adjustments may be required in many modern applications of machine learning studies. In the periodic studies of machine learning applications, not only the model development process is affected, but also the model deployment process may change. In machine learning projects, in addition to the model development phase, the real-time model deployment process also has a very important place. During the model deployment process, continuous integration is attempted by taking model, business, regulatory and flow needs as reference. During the model deployment phase, the integration of the trained model into the decision processes or production environment can be achieved through quantization, memory allocation, mapping and scheduling. During the model deployment process, real-time systems can be developed with front-end, back-end and middleware architecture. In fact, recently, a different dimension has been added to model deployment processes with cloud computing and cloud platform technology. Despite all the positive developments in the model deployment process, perfection can be quite difficult in model integration. Although the model deployment process is time-consuming and error-prone, it is also open to recent developments and interactions. In the first stage of this paper, a double deep q network model was developed as an alternative to traditional classification studies for card fraud detection. In this study, it was concluded that the double deep q network model is quite successful compared to traditional classification models. In addition, the process integration and deployment phase of the champion card fraud model is also discussed in study.