Two-stage fast classifier based on linear classification tree and neural network

Abstract

The invention discloses a two-stage fast classifier based on a linear classification tree and a neural network. Design of the classifier is fundamental and critical in machine learning and pattern recognition, and the classifier is widely applied to numerous fields of data mining, data analysis, expert systems, biomedicine, agriculture and the like. The classifier achieves splitting and recognition of normal massive various sample sets with distinguished features. In normal conditions, the sample sets are approximately divided into linearly separable and linearly inseparable classes. Firstly, the linearly separable sample sets are classified, namely the classes distinguished in features are classified by related statistical knowledge to form the linear classification tree; secondly, relatively unimportant samples are deleted in order to increase correct resolution ratio of the neural network; and thirdly, since the neural network has strong approximation capability and generalization capability, the nonlinear classifier is the classifier based on the neural network. Aiming at long training time of the neural network, scale of the samples is decreased by the linear classification tree and the dimension reduction technology. Besides, the problem of unstable output of the neural network is solved by adjusting objective function of the neural network and verifying whether training indexes of the neural network meet requirements or not.

Description

technical field [0001] The invention relates to a fast classifier for machine learning and pattern recognition, and is especially aimed at the problems of normal, obvious feature and massive data set segmentation and recognition. A fast classifier includes two important stages: constructing a linear classification tree and constructing a neural network. The linear classification tree mainly classifies the samples with obvious characteristics, which can reduce the time of the nonlinear classifier and improve the generalization ability of the nonlinear classifier; the neural network mainly classifies the sample sets with overlapping. Background technique [0002] The design of classifiers is one of the most basic and critical issues in machine learning and pattern recognition. It is widely used in machine learning, pattern recognition, data mining, data analysis, intelligent systems, expert systems, biomedicine, agriculture and other fields. Not only has broad application pro...

AI Technical Summary

Problems solved by technology

[0008] In the actual application process of neural network, there are the following disadvantages: (1) The training time of neural network is long, that is, it takes a long time for neural network to adjust the appropriate weight from the initial weight

Several important factors leading to this reason include: more training data; lower efficiency of training algorithm; too many adjustment parameters in the network

(2) The instability of the neural network, that is, the results of each training of the neural network are different, sometimes the effect is good, sometimes the effect is not good, mainly caused by the algorithm that adjusts the network weight

The disadvantages are: (1) It cannot handle the classification problem with cross sample sets, that is, the classification tree cannot handle linear inseparable sample sets; (2) The boundary determination problem of the classification tree is difficult to solve, that is, the judgment conditions of the intermediate nodes are difficult to determine

Its disadvantages are: (1) For large-scale training samples, SVM requires a lot of training time; (2) because support vector machines can only deal with binary classification problems, this method cannot solve multi-classification problems well

[0014] To sum up, for large-scale classification problems, the existing classification methods mainly have long training time, unstable output results, low training accuracy, and multi-category classification problems.

Images