articles

Methods for knowledge testing are discussed and a new method for testing results refinement is suggested. The method is based on a combination of traditional testing methods and models of neural networks model. The methods proposed were the basis of the developed application software.

Comparative analysis of popular image compression methods based on discrete cosine transform and fractal coding is presented. Image compression principles incorporating these approaches are discussed. The obtained results make it clear the promise of fractal coding technique implementation for image compressing tasks.

Computer system mathematical model is suggested. Program execution process is represented as a process of OS resources consumption which are distributed among consumers. The task of estimating backup progress and execution time using the approach is described. The experimental data show good affinity to the theoretical ones.

The paper presents high-precision software techniques for random distributions number generating. Generally the generators are embedded in simulation systems used to analyze the processes occurring in risk environment. Some of the tools are introduced for the first time.

The method for reconfiguration and training of RBF network parameters adjusting is presented. An approach suggested is based on determining the quality for current BF functions covering of training set and their dynamical recalculation. The neural model features high performance and enables automated radial functions parameters evaluation. Both RBF network development and the network optimization applications are allowed.