Search

Outlier detection has important applications in various data mining domains such as fraud detection, intrusion detection, customers' behavior and employees' performance analysis. Outliers are characterized by being significantly or "interestingly" different from the rest of the data. In this paper, a novel cluster-based outlier detection method is proposed using a humoral-mediated clustering algorithm (HAIS) based on concepts of antibody secretion in natural immune systems. The proposed method finds meaningful clusters as well as outliers simultaneously. This is an iterative approach where only clusters above threshold (larger sized clusters) are carried forward to the next cycle of cluster formation while removing small sized clusters. This paper also demonstrates through experimental results that the mere existence of outliers severely affects the clustering outcome, and removing those outliers can result in better clustering solutions. The feasibility of the method is demonstrated through simulated datasets, current datasets from the literature as well as a real-world doctors' performance evaluation dataset where the task is to identify potentially under-performing doctors. The results indicate that HAIS has capabilities of detecting single point as well as cluster-based outliers. [ABSTRACT FROM AUTHOR]

Nature over millions of years has found innovative, robust and effective methods through evolution for helping organisms deal with the challenges they face when attempting to survive in hostile and uncertain environments. Two critical natural mechanisms in this evolutionary process are variation and selection, which form the basis of "evolutionary computing" (EC). EC has proved successful when dealing with complex problems, such as classification, clustering and optimization. In recent years, as our knowledge of microbiology has deepened, researchers have turned to micro-level biology for inspiration to help solve complex problems. This paper describes a novel supervised learning algorithm inspired by the humoral mediated response triggered by the adaptive immune system. The proposed algorithm uses core immune system concepts such as memory cells, plasma cells and B-cells as well as parameters and processes inspired by our knowledge of the microbiology of immune systems, such as negative clonal selection and affinity thresholds. In particular, we show how local and global similarity based measures based on affinity threshold can help to avoid over-fitting data. The novelty of the proposed algorithm is discussed in the context of existing immune system-based supervised learning algorithms. The performance of the proposed algorithm is tested on well-known benchmarked real world datasets and the results indicate performance not worse than existing techniques in most cases and improvement over previously reported results in some. The role of memory cells is highlighted as a key feature in AIS-based supervised learning that deserves further exploration and evaluation. [ABSTRACT FROM AUTHOR]