@article{2363,
  author = {Hongxing Li, Wenjin Li, Jingrong Shu, Chen Ye-Hui},
  title = {Research on Detection Algorithm of Multi dimensional Outlier Based on Weighted Entropy and BP Neural Network},
  journal = {Signals and Telecommunication Journal},
  year = {2017},
  volume = {7},
  number = {2},
  doi = {},
  url = {http://www.dline.info/stj/fulltext/v6n2/stjv6n2_2.pdf},
  abstract = {At present, the accuracy and robustness of unsupervised learning algorithm for multidimensional outlier
detection are not good, this paper conducts a supervised learning for partial data by supervised BP neural network.
Reliability of data is guaranteed by weighted entropy, then the author learns the relationship between non outliers and
outliers through BP neural network, forecasts test data by trained BP neural network, finally compares the change of the
data position before and after the prediction, take it as multidimensional data outlier distance, so as to complete this test.
This article carries on the test experiment on the data set of multiple data sets. Compared with the traditional LOF and PSO
methods, the experimental results show that the robustness and accuracy of the proposed algorithm for outlier detection of
multidimensional data are better, which is suitable for all kinds of outlier detection of multidimensional data.},
}