The speed prediction algorithm introduced in this paper takes advantage of fuzzy systems that are insensitive to random noise, robust to uncertainties, and transparent to interpretation. The proposed algorithm for outlier detection selects the potential outliers based on the density rather than the deviation adopted in conventional approaches. To evaluate the developed system, a seris of experiments conducted on the real world data. The result of the comparison performed to evaluate the outliler detection method proposed reveals the benefit from the consideration of density. The cross validation results indicate the effectiveness of the fuzzy inference system developed.
|Title of host publication||Intelligent Robotics and Applications: First International Conference, ICIRA 2008 Wuhan, China, October 15-17, 2008 Proceedings, Part I|
|Editors||C. Xiong, Honghai Liu, Y. Huang, Y. Xiong|
|Place of Publication||Berlin|
|Number of pages||10|
|Publication status||Published - 2008|
|Name||Lecture Notes in Computer Science|