A survey of anomaly detection techniques

The phrase "anomaly detection" is often used to describe any technique that looks for samples that differ from expected patterns. Depending on availability of data labels, types of abnormalities and applications, many anomaly detection techniques have been developed. This study aims to give a well-organized and a thorough review of anomaly detection techniques. We think it will aid in a better understanding of the topic of anomaly detection. It also presents the different approaches introduced in the literature for anomaly detection from images as well as other patterns. Despite the common availability of categorical data in practice, anomaly detection from categorical data has received a relatively little attention as compared to that from quantitative data. We divide the anomaly detection research methodologies into distinct categories. We describe the fundamental anomaly detection techniques, as well as their modifications and importantance. In addition, we highlight the merits and demerits of each category. Finally, we discuss the research gaps and limitations encountered, when using anomaly detection techniques for categorical data to solve real-world problems.