Wyniki wyszukiwania - BazTech

1

Recent progress in ant algorithms for fixed telecommunication networks: a review

Gadomska M., Pacut A.

Prace Naukowe Politechniki Warszawskiej. Elektronika

|

2007

|

z. 160

79-89

EN

The aim of this paper is to provide a comprehensive review of Swarm Intelligence based methods for fixed packet-switched telecommunication networks. Such methods, thank to their distributed form and self-organizing dynamics, are very well applicable to problems encountered in a network environment. we believe to show the motivation for using learning methods to solve the routing problem and moreover to create a basis for researchers who act in the field of Swarm Intelligence.

2

Programowy moduł analizatora Zmienności Rytmu Serca (HRV) - opis funkcjonalny oraz algorytmiczny

Piotrowski J., Różanowski K.

Prace Naukowe Politechniki Warszawskiej. Elektronika

|

2007

|

z. 161

23-30

EN

The paper presents HRV tool v.1.0 software for Heart Rate Variability analysis. This software is developed in Matlab environment to verify various HRV spectrum estimation methods. HRV tool software is not dedicated for clinical use and it is a software demonstrator rather than commercial use software. HRV parameters estimation is based on advanced spectral analysis of electrocardiogram ( tachogram of RR values). HRV is noninvasive method of estimation the Sympathetic and Parasymathetic Nervous System influence on the heart rate. HRV spectral analysis is performed based on Power Spectral Density estimation using parametric, nonparametric as well as spatial methods. Standard parameters of HRV spectral analysis are computed e.g.: TP, ULF, VLF, LF, HF, LF/HF, frLF, frHF, alpha for dedicated region of interest.

3

Performance comparison of the AS and GTSalgorithms for weighted maximum leaf spanning tree problem

Czaderna P., Wala K.

Prace Naukowe Politechniki Warszawskiej. Elektronika

|

2007

|

z. 160

43-50

EN

In the paper the combinatorial model of the weighted maximum leaf spanning tree problem of simple graph is presented. We give a detailed description of three procedures of the neighbour trees generation to the basis one which are exploited during the course of tree improving process realized by the classical simulated annealing and genetic local search algorithm. Numerical results of improving processes for randomly generated praphs are included.

4

On two models for global optimization

Żilinskas A.

Prace Naukowe Politechniki Warszawskiej. Elektronika

|

2007

|

z. 160

9-16

EN

Two models for global optimization are considered: statistical model, and radial basis functions. The equivalence of both models in the case of optimization without noise is discussed. Both models are also evaluated with respect to global optimization in the presence of noise by means of experimental testing where approximation errors of passive one dimensional algorithm are estimated.

5

Multiobjective optimization of heat radiators using evolutionary algorithms

Długosz A., Burczyński T.

Prace Naukowe Politechniki Warszawskiej. Elektronika

|

2007

|

z. 160

51-58

EN

The paper is devoted to the multiobjective optimization of heat radiators using eveolutionary algorithms. The proposed algorithm, based on the Pareto approach, is applied in optimization of heat radiators used to dissipate heat from electrical devices. The boundary-value problem for thermoelasticity is solved using the finite element method. Numerical examples are also included in the paper.

6

Knowledge reuse for an ensemble of GP-based learners

Jaśkowski W., Krawiec K., Wieloch B.

Prace Naukowe Politechniki Warszawskiej. Elektronika

|

2007

|

z. 160

135-142

EN

We propose a method of knowledge reuse for an ensemble of genetic programming-based learners solving a visual learning task. First, we introduce a visual learning method that uses genetic programming individuals to represent hypotheses. Individuals-hypotheses process image representation composed of visual primitives derived from given training images that contain objects to be recognized. The process of recognition is generative, i.e. an individual is suposed to restore the shape of the processed object by drawing its reproduction on a separate canvas. This canonical method is in the following extended with a knowledge reuse mechanism that allows a learner to import genetic material from hypotheses that evolved for other decision classes (object classes). We compare the performance of the extended approach to the basis method on a real-world tasks of handwritten character recognition, and conclude that knowledge reuse leads to significant convergence speedup and reduces the risk of overfitting.

7

DNA sequence analysis

Biedrzycki R.

Prace Naukowe Politechniki Warszawskiej. Elektronika

|

2007

|

z. 160

35-42

EN

This paper presents a brief survey of computational approaches to the DNA sequence analysis. The basic biological background is presented. The various types of algorithms for pattern construction and gene finding are presented with special attention paid to the application of global optimization methods.

8

Combining negative selection with immune K-means algorithm for improving the support vector machines method

Bereta M., Burczyński T.

Prace Naukowe Politechniki Warszawskiej. Elektronika

|

2007

|

z. 160

25-33

EN

This paper presents a novel method of using the ideas from Artificial Immune Systems for improving the performance of the support Vector Machines. By means of Immune K-Means algorithm a set of artificial data is generated based on the oryginal training data. The artificial data describes the most important information from the classifiers learning point of view - the information about the boundaries among the classes remain in the artificial data. Combining the Immune K-Means algorithm with Negative Selection methods allows for further improvements of the artificial data set. The proposed approach allows to speed up the learning process of SVM when the training data set is large by extracting the most important information first. The proposed method can also be used as a data compression, especially suited when the information about boundaries among classes is an important issue. The artificial data can be created once and then used for parameters tuning of different classification methods, speeding up the learning process.

9

Application of hybrid evolutionary algorithm to single source capacitated warehouse location problem

Jastrzębski W.

Prace Naukowe Politechniki Warszawskiej. Elektronika

|

2007

|

z. 160

125-133

EN

This paper presents two-phase hybrid evolutionary algorithm (EA) to optimize Single Souece Capacitated Warehouse Location Problem (SSCWLP), a well-known location-allocation problem employed for the telecommunication network design modeling. The first phase of the algorithm aims the search to satisfy the problem constraints and during the second phase actual optimization take place. To improve the performance of EA the algorithm is combined with other local search heuristics. Influence of EA hybridization as well as different selection schemes dealing with constraint handling are discussed. In adition, performance of co-evolutionary algorithm (co-EA) versus the EA with single population is compared across a set of example problems.

10

An application of Bezier curves in the off-line verification of handwritten signatures

Gawor P., Kołodziej J.

Prace Naukowe Politechniki Warszawskiej. Elektronika

|

2007

|

z. 160

99-106

EN

The task of veryfing human signatures as a pattern recognition problem is considered in this paper. We propose two-step vectorization of the binary handwritten signature images with the application of the cubic Bezier curves. We introduce a new contour segments extraction algorithm to generate discrete sets of signature contour points. A multi-layer neural network trained by the real-coded genetic algorithm was proposed as the signature patterns classifier. Some simple numerical tests results are also reported.

11

Adaptation Cost: what is the Repayment Time for the Evolutionary Saddle Crossings?

Galar R., Karcz-Dulęba I., Łojek R.

Prace Naukowe Politechniki Warszawskiej. Elektronika

|

2007

|

z. 160

91-98

EN

It is usual to asses alternative optimization search algorithms in terms of the number of trials necessary to approach the optimum. In case of adaptive algorithms implemented in the real systems such assesment seems inadequate, as the loss of quality due to the exploration of the vicinity of the identified optimum should be also taken into account. We try to evaluate this aspect of adaptive search in the case of a path dependent evolution with proportional selection and normally distributed mutations, which involves saddle crossing in a bimodal adaptative landscape.

12

3-D surface reconstruction by the means of evolutionary algorithms

Goiński A., Obuchowicz A.

Prace Naukowe Politechniki Warszawskiej. Elektronika

|

2007

|

z. 160

107-114

EN

In the paper we propose a new method for an isosurface construction from unorganized points in 3-D. The aim is to combine evolutionary algorithms with a recursive subdivision scheme. the algorithm starts from a base population where each individual represents an N-level surface approximation of an input object. next the population is evaluated and an offspring generation is created with a denser subdivision of surface. The method is illustrated by a set of samples.