Intrusion detection has become a popular solution of mitigating cyber-attacks in this technological era. Deep learning algorithm have become the go to solution in developing intelligent intrusion detection models and they have been successful. However, deep learning algorithms are too slow in training and this may not be effective for real time applications. Therefore, we propose to develop a faster intrusion detection model using a novel majority voting ensemble with Random Forest recursive feature elimination to improve computation time as well as the detection rate. By including the random forest recursive feature elimination in the proposed model, redundant features will be removed while maintaining the most discriminative features. We evaluated the efficacy of the proposed model on the CICIDS 2017 dataset. In addition, we compared our model with deep learning algorithms which include Resnet50, LSTM and AE-FCN. The experimental results showed that our proposed model outperformed the other deep learning algorithms in terms of accuracy, precision, recall and computation time.