Advanced Materials Research Vols. 532-533

Abstract: A novel modeling method based on Bayesian inferring for dynamic nonlinear system is proposed in this article. The Bayesian inferring model structure and its training algorithm combined with evolutionary algorithms are first described in which the matrix threshold D parameters are optimized by evolutionary algorithms and the structure of the Bayesian inferring model is updated by the system running data. Then some typical dynamic systems are used for validating the modeling effectiveness of the Bayesian inferring method. And the simulation results are presented and some conclusion on the Bayesian inferring modeling method is described in details at last.

Abstract: In analysis of association rules based on OLAP, the application of Apriori algorithm may need to scan the database frequently and generate massive number of candidate itemsets because of the complex system. According to the weakness, the paper proposes PApriori algorithm based on pretreatment, which reduces the time of scanning database to once, while not directly generate candidate itemsets, raise the efficiency Analysis of association rules based on OLAP efficiently.

Abstract: Discretization of decision table is the important step for pretreatment of data mining and machine learning, which related to the effect of learning. It has great contribution to speeding up the followed learning algorithms, cutting down the real demand of algorithms on running space and time. In this paper, the basic characteristics and framework of discretization approaches about greedy and improved algorithm are analyzed at first, then a new algorithm is put forward to select the useful cuts. The example is given to show that the useful cuts is consistent with the result of technicist. The algorithm offered the important theoretics basis for followed attribute reduction.

Abstract: The genetic algorithm is a powerful global search and optimization technique based on the principles of natural selection and genetics, but it is not suitable in solving large-scale and complicated problems due to its the shortcomings in computational accuracy and efficiency. Against these deficiencies, a coarse-grained parallel genetic algorithm (PGA) model based on distributed cluster system is proposed in this paper. Flow chart about the model is designed and detailed analysis of migration scheme is offered. This paper investigates the parallel efficiency of the coarse-grained PGA and migration operator by experiments on a specific inverse heat conduction problem .The experimental results show that the model can achieve upper speedup rations, improve computational efficiency and the overall performance of the PGA.

Abstract: Path planning for dual aircrafts is preceded by artificial immune algorithm in forest fire rescue. We analysis and compare several parameters which will influence path planning. Based on above results we sum the reasonable parameters which should be used in further experiment. Moreover, to improve the operability of aircraft, restrictions of gradient smoothing algorithm, curvature smoothing algorithm are performed to smooth track in vertical aspect. Experimental results show that we can successfully get the satisfied flight paths for dual aircrafts in fire rescue under complex circumvent. At the same time the paths we planned have certain practical value by consideration the aircrafts fly-ability constraints.

Abstract: For the purpose of overcoming the premature property and low execution efficiency of the Particle Swarm Optimization (PSO) algorithm, this paper presents a particle swarm optimization algorithm based on the pattern search. In this algorithm， personal and global optimum particles are chosen in every iteration by a probability. Then, local optimization will be performed by the pattern search and then the original individuals will be replaced. The strong local search function of the pattern search provides an effective mechanism for the PSO algorithm to escape from the local optimum, which avoids prematurity of the algorithm. Simulation shows that this algorithm features a stronger function of global search than conventional PSO, so that the optimization process can be improved remarkably.

Abstract: In order to improve the stability and effectiveness of the key frame extraction, we devise a key frame extraction algorithm based on frame image block. It get the local characteristics of the image frame information by partitioning frame image, then calculate the non-correlation coefficient, and extract key frames reflecting more diversified information. Experimental results show that the algorithm for the lens of key frames of extraction efficiency is higher and extraction of key frames can effectively reflect lens content.

Abstract: With the availability of inexpensive storage and the progress in data collection tools, many organizations have created large databases of business and scientific data, which create an imminent need and great opportunities for mining interesting knowledge from data.Mining association rules is an important topic in the data mining research. In the paper, research mining frequent itemsets algorithm based on recognizable matrix and mining association rules algorithm based on improved measure system, the above method is used to mine association rules to the students’ data table under Visual FoxPro 6.0.

Abstract: For the weakness of conventional POCS algorithms, a novel spatio-temporal adaptive super-resolution reconstruction algorithm of video is proposed in this paper. The spatio-temporal adaptive mechanism, which is based on POCS super-resolution reconstruction algorithm, can effectively prevent reconstructed image from the influence of inaccuracy of motion information and avoid the impact of noise amplification, which exist in using conventional POCS algorithms to reconstruct image sequences in dramatic motion. Experimental results show that the spatio-temporal adaptive algorithm not only effectively alleviate amplification noise but is better than the traditional POCS algorithms in signal to noise ration.

Abstract: This paper presents an improved SPRINT algorithm. The original SPRINT algorithm is a scalable and parallelizable decision tree algorithm, which is a popular algorithm in data mining and machine learning communities. To improve the algorithm's efficiency, we propose an improved algorithm. Firstly, we select the splitting attributes and obtain the best splitting attribute from them by computing the information gain ratio of each attribute. After that, we calculate the best splitting point of the best splitting attribute. Since it avoids a lot of calculations of other attributes, the improved algorithm can effectively reduce the computation.