The Structure Elucidation Of Infrared Spectrum By Using Svm

Data-based machine learning covers a wide range of topics from pattern recognition to function regression and density estimation.Most of the existing methods are based on traditional statistics.So they may not work in practical cases of limited samples. SLT is a small-sample statistics by Vapnik which concerns mainly the statistics principles when samples are limited, especially the properties of learning procedure in such cases. The work has been done in this paper as follows:To design a SVM algorithm software named Easysvm .The "one against all" method was proposed on the basis of two-class pattern recognition. This method made the multi-class problems simple and convenient. Easysvm software multi-class algorithm was applied to the functiuonal groups prediction from infrared spectra, the average prediction accuracy was up to 96%, even for some functional groups, the prediction accuracy reached 100%. The model can predict the molecular component of the compound through the infrared spectra data. The result showed it is feasible to use the SVM to distinguish substance, the purity detection and etc.Easysvm software algorithms was applied to the canser data compared with KNN, ANN. The prediction accuracy reached 100%, better than KNN, ANN algorithm use to this problem.Use weight support vector machine algorithms to G-protein coupled receptor, experimentation shows that weight support vector machine can improve the classification accuracy with small size at cost of accuracy of large size class. Theprediction accuracy for positive class reached 80%,the prediction accuracy for negative class is 95.45%.The experimental results reveal that Easysvm software is stable and easy to use in the practice. When solving the problem to classify the pattern, there will be within two thousand samples, and the programing from the software can quickly and safely find the optimum solution. In the practice of concrete mathematic modeling, Easysvm software show good application, steadily solving many problems in study and practical application, and is widely used in many researches of chemical cic laboratory, so Easysvm software is available to lab research and simple, practical tool software.