The Model for Estimation of Computer System Used Resources While Extracting Production Rules Based on Parallel Computations

Abstract

Context. The task of production rules extraction while processing big arrays of data has been discussed. The problem of estimation of computer system used resources while extracting production rules based on parallel computations has been solved. The research object is the process of production rules extraction. The research subject lies in methods of parallel computer systems' resource planning.Objective. The purpose of the work is а construction of the model for estimation parallel computer systems resources used to solve applied problems based on the parallel method of production rules extraction.Method. The article deals with the model building of used resources estimation of parallel computer system while extracting production rules. The model for estimation of computer system used resources while executing the parallel method of method of production rules extraction is proposed. Synthesized model takes into account the type of computer system, the amount of processors involved to solving the task and the bandwidth of data transfer network. In addition, the model considers parameters of used mathematical equipment (the portions of parallel system nodes involved for production rules extraction based on decision trees, associative rules and negative selection). Also the parameters of solved application task are taken into account. They are the number of observations and the number of characteristics in a given set of data describing the results of observations of the object or process being studied. The synthesized neural model is a polyalgorithmic. It allows estimating two characteristics of parallel computer system while executing the parallel method of production rules extraction. The first one is time used. And the second one is the volume of memory used.Results. The software which implements the proposed model and allows predicting the time and the volume of memory used of parallel computer system while solving practice tasks has been developed.Conclusions. The conducted experiments have confirmed the proposed software operability and allow recommending it for use in practice for solving the problems of big data processing. The prospects for further research may include the creation of parallel methods for feature selection, as well as an experimental study of proposed model on more complex practical problems of different nature and dimensionality.

Development of stratified approach to software defined networks simulation

“…The adequacy of the resulting coupled DEVS-model has been proven with t-criterion for degrees of freedom 8 The advantages of the proposed approach are conditioned by the fact that it is aimed solely at simulation (not taking into consideration the emulation aspects). Here are some of these: network visualization-related labor costs lowering -no need to deploy Xming X Server and use PuTTY utility, working on widespread Windows-platform; easiness of system model reconfiguration.…”

“…Shevchuk, V. P. Metrologiya intellektual'nyh izmeritel'nyh sistem [Text]: monografiya / V. P. Shevchuk, V. I. Kaplya [5][6][7], Big Data [8], Artificial Intelligence [9], Internet of Things [10]. For instance, the synergy of SDN and Neural Networks technologies can be expedient in the context of parallel computing systems' resources allocation [11,12].…”