The management of future networks is expected to fully exploit cognitive capabilities that embrace knowledge and intelligence, increasing the degree of automation, making the network more self-autonomous and enabling a personalized user experience. In this context, this paper presents the use of knowledge-based capabilities through a specific lab experiment focused on the Channel Selection functionality for Cognitive Radio Networks (CRN). The selection is based on a supervised classification that allows estimating the number of interfering sources existing in a given frequency channel. Four different classifiers are considered, namely decision tree, neural net-work, naive Bayes and Support Vector Machine (SVM). Additionally, a comparison against other channel selection strategies using Q-learning and game theory has also been performed. Results obtained in an illustrative and realistic test scenario have revealed that all the strategies allow identifying an optimum solution. However, the time to converge to this solution can be up to 27 times higher according to the algorithm selected.