A dynamic data correction algorithm based on polynomial smooth support vector machine

• Published in: 2016 International Conference on Machine Learning and Cybernetics (ICMLC) Vol. 2; pp. 820 - 824
• Main Authors: Dong-Mei Pu, Da-Qi Gao, Yu-Bo Yuan
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.07.2016
• Subjects: Aerodynamics; Big data; Cybernetics; Data analysis; Data correction; Data Mining; Heuristic algorithms; Machine learning algorithms; Support vector machine; Support vector machines
• ISSN: 2160-1348
• DOI: 10.1109/ICMLC.2016.7872993

Data quality plays an important role in modern intelligent information system and is crucial to any data analysis task. Many imperfection-handling techniques avoid overfitting or simply remove offending portions of the data. Data correction can help to retain and recover as much information as possible from the original data resources. In this paper, we proposed a novel technique based on polynomial smooth support vector machine. The quadratic polynomial and the first degree of polynomial as the support vector machine smooth functions are investigated. At the same time, the function was used as smooth function to calculate compensation values. In order to show the procedures of our algorithm, some necessary steps need to be considered. Firstly, the original data are normalized, so as to eliminate experimental effects of dimensional problems. Secondly, the three different kinds of smooth functions need to be analysed mathematically. The difference measure are calculated to make sure the results of correction through different data correction models. The results of given noised data sets can show that the proposed the data correction method based on polynomial smooth support vector machine is effectiveness.