When studying the military leaders' behavior, we observe that the problem of prediction transformed their work in a Sisyphus one. The lack of knowing the future, especially in the field of states' behavior, represents a major problem for the military decision makers. In this article one will not use the prediction methods to generate possible futures, but will use it in order to analyze if the predictions methods are still a living concept. We asked whether the use of these methods could be a solution with encouraging results in developing military leaders competencies. In order to validate this hypothesis one tried to bind the decision types with neural networks and with military leaders behavior, thus creating a holist course consisting of variables united by a fitting problem. These elements could indicate one through a scientific approach if the prediction methods are really important for nowadays society. One used some methods which are widely spread in the international academic community in order to obtain the leaders decisions. Of course, it was shown that the behavior of the military leaders is connected with the approval of some prediction methods. We argue that the state of security is described by some patterns. These patterns could indicate us, through a scientific approach, if the insecurity is inherent. The paper mainly refers to Tangredi, Kahn and Schwartz analysis from this field. It tries to emphasize the importance of the scenarios and neural networks in security. Finally, the paper tried to emphasize the importance of prediction in the field of security.

This article proposes the use of neural networks as a solution with encouraging results in making predictions in the fields of security and international relations. It tries to bind the Copenhagen Security Model with neural networks, thus creating a holist course consisting of variables united by a fitting tool problem. These elements could indicate us through a scientific approach if the insecurity is inherent. The paper mainly refers to Tangredi, Kahn and Schwartz analysis in this field. One uses some datasets which are widely spread in the academic community (Correlates of war and Polity 4) in order to test the model. The results obtained from Eugene and Matlab Neural Network Toolbox are compared with the results of H. Kahn and B. Russet. Of course, it is shown that flexibility of the neural network models is able to reach important empirical relations drawn between democracy and conflict which aren't seen with other means. Finally, the paper tries to emphasize the importance of the scenarios and neural networks in security. One chooses to agree with supporting the neural networking because their usage opens a new age in the field of predictions. One demonstrates that this fact is happening because the neural model offers us the possibility of learning. It means that as long as the model roles it will correct its errors and become more accurate. It represents a valuable option for future studies which changes the meaning of the scenario from the tool used for developing simple tales of possible futures to description with "full ramifications" which are designed to change and modify our leaders' view of reality.