The Intent-Based Networks (IBN) are believed to be the future of the networking. They incorporate the advantages of many modern networking paradigms. It is believed that they may operate in a globally optimal manner de-coupling the data and control plane as the SDNs. Resource management is conducted with the use of the virtual networks concept. However, in opposite to the classic networks, IBNs are configured automatically, and the objectives and requirements are given by administrators and users as high-level business intents. In this paper, we introduce the idea of the machine learning based management system that enables IBN to operate according to its paradigm. We present the usage of our system to perform virtual network admission control tasks. We present the results of some initial simulations that were conducted to verify our approach. The outcomes of the experiments are very promising.