The current report shows over 1.5 billion desktop computers are connected to the internet worldwide and over 6.9 billion smartphone users worldwide. These digital assets store value, make payments and exchange goods and services. The increasing volume of these digital assets makes them a prime target for cybercriminals. As such, protecting our digital assets from cyber threats is essential. We use IDS as a prominent and versatile tool to secure digital assets. Over the last two decades, researchers have used signature-based IDS. These systems need to be more accurate, and it is difficult to differentiate between malicious and non-malicious activities. As per the current scenario, we need robust intrusion detection systems that are fast and accurate. Conventionally, two mechanisms are used for intrusion detection in compromised systems: behavior-based and knowledge-based. In this paper, the authors explored the possibilities of applying machine learning techniques and hybrid mechanisms for intrusion detection based on signature and anomaly detection methods. (CSE-CIC-IDS2018) dataset is used to check the performance of heterogeneous machine learning classifiers such as KNN, SVM, NB, and AE.