Gu Xiang

A Synthesized Data Mining Algorithm Based on Clustering and Decision Tree

With the development of information technology and computer science, high-capacity data appear in our lives. In order to help people analyzing and digging out useful information, the generation and application of data mining technology seem so significance. Clustering and decision tree are the mostly used methods of data mining. Clustering can be used for describing and decision tree can be applied to analyzing. After combining these two methods effectively, we can reflect data characters and potential rules syllabify. This paper presents a new synthesized data mining algorithm named CA which improves the original methods of CURE and C4.5. CA introduces principle component analysis (PCA), grid partition and parallel processing which can achieve feature reduction and scale reduction for large-scale datasets. This paper applies CA algorithm to maize seed breeding and the results of experiments show that our approach is better than original methods.

Advances in Research of Fuzzy C-Means Clustering Algorithm

Fuzzy Clustering Analysis is one of the most popular research currently. It attracts peoples attention for its validity in resolving fuzzy problem. With the development of computer science, Fuzzy Clustering Analysis based on objective function became hot topics and Fuzzy C-means Clustering Algorithm is one of the most perfect and most widely used theories although there are some drawbacks for the classical algorithm itself. This paper summarizes the progress of FCM algorithm, and some algorithm improvements were analyzed and compared.

An Expanding Products Algorithm of Logic Functions Optimization

This paper presents an expanding products algorithm of logic optimization that is based on multi-input multi-output logic functions. According to input products, we work out complement sets matrix of non-shared products, blocking matrix, covering matrix. For blocking matrix and covering matrix, we get mini column cover sets and upgrade sets based on choice algorithm of mini column cover. Then we form product express after input product that be expanded. For multi-input multi-output logic functions of two-level SOP, we design a program. Max-input variables is 128, max-output variables is 256, sum of max-input-output variables is 300, max-input products is 20000. It overpasses testing of Benchmark.

Improved Master-Slave Parallel Task-Allocating Algorithm and Its Implement

This paper is based on the research of master-slave parallel task allocation, and points out the algorithm not considering the cost of the task which will be added to processor. Considering that would cause the disequilibrium of task allocation on slaving processors, this paper presents the improved master-slave parallel task-allocating algorithm. Before task groups being assigned to slave-processors, the main processor makes all the task groups being in descending order of their cost. Avoiding costly task groups concentrating in the busy processor, this manipulation makes the free processor getting the most costly task group of the rest task groups at all time and achieves the goal of load balance. The experimental results show the advantages of the new algorithm and certify that it efficiently solves the bad balance problems, and improves the efficiency.

The Application of Information Fusion and Extraction in Maize Seed Breeding

It has been the irresistible trend of agriculture informationization which uses information technologies to treat enormous data and finds out potential useful rules to direct the development and reformation of agriculture. Aiming at the specific application of maize seed breeding, this paper effectively integrates several data mining technologies and presents a new method called CA to analyze the whole maize information. The algorithm achieves transverse dimension reduction by combining PCA and other methods, and also achieves longitudinal dimension reduction by improving CURE and k-means. The decision-tree method of CA algorithm introduces three different classifiers in order to enhance the accuracy of trees. By comparing the results of improved algorithm with traditional methods, we can find that the new algorithm is better in performance and degree of parallelism.

Minimization Algorithm of Unate Logic Functions

Unate logic functions is an especially case of logic functions. For logic functions optimization, it costs mor spaces and times, spending of memory is increase on two power by input variables. According by characteristic of unate logic functions, this paper introduces an algorithm of unate logic functions optimization and presents a method of algorithm realization. It analyzes influence factor of unate functions logic optimization efficiency. The software system overpasses the testing of right validate. The results of experimentation is proved to be correct and efficient. It is good for reduce spending of memory spaces and times. Optimizing of effect is reduced by increasing numbers of input variable in case of the same numbers of output variable??? products and don’t-care gene, and by increasing numbers of don’t-care on case of the same numbers of input variable???output variable and products. The optimizing of effect is increased by increasing numbers of products on case of the same numbers of input variable???output variable and don’t-care gene.