Identification and Classification of Fraud Crime by Cooperative Management using Deep Neural Networks

Abstract

Cooperatives are organizations that are legal entities and their development in Indonesia is part of the overall national development effort. Cooperatives must be built to create businesses and services in creating the principle of kinship. The problem studied is how the form of fraud against cooperatives and how criminal liability against cooperatives. In this paper, potential criminal acts by cooperative management are identified and classified using the deep neural networks (DNN) method from the data of 275 respondents. The tests at the data processing stage with the DNN model show that the model with 140 epochs gives better estimation results with an accuracy of 98.5%, while the use of only 40 epochs gives better estimation results with an accuracy of 93.9%. The use of additional features as input to the DNN model shows the estimation results with a lower error value of 0.188%. The need for supervision of fellow cooperative members in running cooperatives as well as the role of the government in supervising cooperatives that have committed fraud crimes that have resulted in restless fellow members and the community. Strengthening the Cooperative Law so that it can directly ensnare cooperative members who abuse the function of authority or commit criminal acts is also very much needed.