Cyberattacks and security breaches in the Internet of Things (IoT) networks are growing recently due to some inherent security flaws. Thus, the protection of the IoT networks and infrastructure is a major research challenge. To identify the cyberattacks in IoT networks, an intrusion detection system (IDS) is deployed at the edge of the IoT network. Most of the extant IDS suffer from the curse of dimensionality problem, which reduces the overall efficiency of the IDS system. Therefore, it is crucial to eliminate redundant and irrelevant features from IoT network traffic to develop an efficient IDS. Inspired by aforesaid challenges, this paper has proposed an efficient IDS model using ensemble feature selection techniques for solving the curse of dimensionality issue. In addition, ensemble classification techniques are applied for the detection and classification of cyberattacks. In the proposed model, first, two distinct feature subsets using filter-based feature selection methods such as Mutual Information (MI) and One-way Analysis of Variance (ANOVA) are identified. Next, these reduced feature sets are combined into a single set using the logical OR operation. As a final step in the detection process, the feature set is passed to two renowned Machine Learning (ML) approaches named Extreme Gradient Boosting (XG-Boost) and Extra Tree (ET). The proposed model is assessed and compared to other recent existing approaches using the BoT-IoT dataset. Based on experimental results, the proposed model is capable of protecting IoT networks against modern cyberattacks with an accuracy rate of 99.99% and a false positive rate of 0.001%.