Overview and contributions -- Developments in unsupervised outlier detection research -- A fast distance-based outlier detection technique using a divisive hierarchical clustering algorithm -- A k-nearest neighbour centroid based outlier detection method -- A minimum spanning tree clustering inspired outlier detection technique -- A k-nearest neighbour spectral clustering based outlier detection technique -- Enhancing outlier detection by filtering out core points and border points -- An effective boundary point detection algorithm via k-nearest neighbours based centroid -- A nearest neighbour classifier based automated on-line novel visual percept detection method -- Unsupervised fraud detection in environmental time series data.
Summary:
"This book enriches unsupervised outlier detection research by proposing several new distance-based and density-based outlier scores in a k-nearest neighbors setting. The respective chapters highlight the latest developments in k-nearest neighbor-based outlier detection research and cover such topics as our present understanding of unsupervised outlier detection in general; distance-based and density-based outlier detection in particular; and the applications of the latest findings to boundary point detection and novel object detection. The book also offers a new perspective on bridging the gap between k-nearest neighbor-based outlier detection and clustering-based outlier detection, laying the groundwork for future advances in unsupervised outlier detection research. The authors hope the algorithms and applications proposed here will serve as valuable resources for outlier detection researchers for years to come.--back cover.
This resource is supported by the Institute of Museum and Library Services under the provisions of the Library Services and Technology Act as administered by State Library of Iowa.