Multi-classifier Combination And Its Appliance In The Lithology Identify

The objective of multi-classifier combination is to make use of each classifier's good qualities in the recognition performance and gains higher recognition rate than each classifier. Multi-classifier combination improves the recognition rate with complementarities among these classifiers. In the traditionary combination methods there are two disadvantages. One is that to select the classification determine their nature, don't combine with the discriminating target. Shortcoming that do it in this way is not to be good full play strong point of very classifier. Two is the role of each classifier in combination is fixed. In other words, to all samples we adopt only one combination method. This method can improve recognition rate for some samples but reduce it for others. To realize the best performance for almost all the samples we need to choose some classifiers and to adopt different combination to the different sort of samples.The text sums up the study headway of the classifier combination in the domestic and foreign and summarizes the reason of classifier combination and the combination method. First the text recommend some pick-up technique of the index and we comparatively study the motheds. Secondly, make simple summary to the reason followed from forefathers why the classifier combination can improve recognition rate for sample. According to this text of reason the text bring forward a rule of choosing classifiers: the wrong relativity of classifiers should try hard little. Then we set up a model to choose the classifier on the basis of the discerned target—Centralized decision method of the fuzzy opinion. Later the text set up a model under the jurisdiction of one fuzzily degree of functions to the classifier combination based on to the chosen classifiers. Consider the classifier combination not only relates to the performance of the classifiers, but alse relates to the sorts of the discerned target. The mothed to make weight up in the paper is ask by criterion classification device under every classification optimum according to the MSE rule and the different sorts of the discerned target. Finally, took an example of the literary discernment of the qiaokou oil gas field of Shengli Oil Field to prove the method. The result shows the model of the text can improve the rate of recognition. So the method is feasible and effective. The text introduce the shortcoming of the method and give out the constructive advice in the end.