The shortage of available communication spectrum bands caused by the increase in demandfrom wireless communication systems is known as the spectrum scarcity problem. Cognitive radios are a potential solution for this problem. By dynamically sensing the availability of spectrum, cognitive radios are able to autonomously adapt their transmission parameters to exploit unoccupied spectrums. Adversarial jamming is a major concern for wireless communication systems as jammers can significantly degrade communication performance. Anti-jamming systems require highly accurate sensing of jammers. In this context, the following two types of jamming detection algorithms for a wide-band cognitive radio were evaluated and compared in this paper: Artificial Neural Network (ANN) and ensemble tree. The proposed algorithms attempt to classify the signals as one of 5 jammer scenario types using extracted cyclostationary features, through the spectral correlation function (SCF). Overall, the simulated performance for both algorithms were shown to have classification accuracies of over 95%. Although classification accuracies of both algorithms were similar, the ensemble tree method is recommended for its shorter training time.