Mining with Noise Knowledge: Error Aware Data Mining

Real-world data are dirty, and therefore, noise handling is a defining characteristic for data mining research and applications. This talk will review existing research efforts on data cleansing and classifier ensembling in dealing with random noise, and then present our recent research on an error aware data mining design to process structured noise. This error aware data mining framework makes use of error information (such as noise level, noise distribution, and data corruption rules) to improve data mining results. Experimental comparisons on real-world datasets will demonstrate the effectiveness of this design.