Self-learning strategies in multiagent environments

A method for revealing and resolving conflicts is presented, especially well applicable for resolution of contradictions. It is shown that agents receive a greater autonomy via a correctly directed or selectable identification and conflict resolution inside the accumulated knowledge. The advantages of the introduced method are presented compared to artificial neural networks (ANN) and other trainable tools with or without a teacher. The offered material is tightly bound to the research presented at former conferences IS´02 and IS´08. Applications are oriented mainly to serve the goals of information security but for the sake of brevity their descriptions cross the borders of the present paper. The introduced method is no less successive in applications for abstract and applied mathematics, in neuro-fuzzy systems, etc. The method may be efficiently combined with other well-known methods and technologies: ANN, machine learning, statistical learning and data mining, knowledge discovery etc. In case of combined exploitation with ANNs a `critical learner´ may be constructed who should establish an active feedback with the teacher and make a deep learning.