支持向量机的发展（Support vector machine development）.docVIP

支持向量机的发展（Support vector machine development） The long taiyuan to hide the problem, to mourn the difficulties of the peoples livelihood. The window contains the west ridge snow, door mooring east wu wanli boat. Step by step, read carefully. It is advisable to go after the poor, not to be a good student wang. East of the dajiang river, the waves have gone through, the ancient and romantic characters. Support vector machine development Since the classic SVM in the early 1990s, due to its complete theoretical framework and many good effects in practical application, it has received a lot of attention in the field of machine learning. Its theory and application have developed both horizontally and longitudinally. Theory: 1. The fuzzy support vector machine, the introduction of samples of category membership functions, so that each sample for the category is different, the effects of the application of this theory to improve the anti-noise ability of SVM, especially suitable for failing to fully reveal the characteristic of input sample. 2. Least squares support vector machine. This approach was proposed in 1999 and has been used in many related fields over the years. The problem of research has been extended to: the processing of large-scale data sets; Robustness of data processing; Parameter adjustment and selection problems; Training and simulation. 3. Weighted support vector machines (weighted with biased samples, biased risk weighting). 4. Support vector machine for active learning. Active learning can be used to further train the classifier based on the learning process and select the samples which are most favorable for classifier performance, and the number of samples can be effectively reduced. In other words, the effectiveness of the classification is sorted by some standard, then the effective sample is selected to train the support vector machine. Rough sets combined with support vector machines. Firstly, the attribute of data is reduced by rough set theory,