We found a match
Your institution may have rights to this item. Sign in to continue.
- Title
An enhanced local outlier detection using random walk on grid information graph.
- Authors
She, Chunyan; Zeng, Shaohua
- Abstract
Outlier detection is a hot issue in data mining, which has plenty of practical applications. Local Outlier Factor algorithm, as a pioneering work of local outlier detection, has been paid much attention. However, it needs to perform the neighbor search with high time complexity and ignores the local distribution of an object within its neighbor. In this work, a novel local outlier detection method based on grid random walk is proposed, which uses random walk to obtain stationary distribution vector of grid information graph. Some grids with small values of stationary distribution vector will be considered as candidate outliers. The outlier detection is performed only on candidate outliers to improve the running efficiency. Then, considering the local distribution of an object within its neighbor, a new local outlier factor is constructed to estimate the abnormal degree of each object. The experimental results indicate that the proposed algorithm has better performance and lower running time than the others.
- Subjects
RANDOM walks; OUTLIER detection; DATA mining
- Publication
Journal of Supercomputing, 2022, Vol 78, Issue 12, p14530
- ISSN
0920-8542
- Publication type
Article
- DOI
10.1007/s11227-022-04459-7