Peter Sincak

Hierarchical fuzzy inference system for robotic pursuit evasion task

We propose hierarchical multi agent control system based on rule based fuzzy system for pursuit-evasion task and state a new representation of this type of game that is based on fuzzy logic. This approach enables improvement of the rule base under uncertain conditions and can process a priori inserted expert knowledge. Example application domain includes reckon and guard robots, research space probes, coordination of multiple mine sweeping devices or autonomous rescue teams.

Performance-based adaptive fuzzy control of aircrafts

Aircraft behavior can be described by a set of parameters that represent their aerodynamic properties and flight conditions. They are measured in aerodynamic tunnels under some special laboratory conditions. However, they do not fully describe all possible flight situations. Their accuracy is also limited. As considerable changes of the description occur during a flight, the use of non-adaptive autopilots, especially in combat aircraft, is strongly limited. The removal of this obstacle is possible only by means of a continuous on-line adaptation of the aircraft model and by a consecutive adaptation of the autopilot. The paper deals with the design of a performance-adaptive, so-called self-organizing, fuzzy controller as an autopilot. The structure of such a controller as well as the modified adaptation principle with some experiments are described.

Combining Gradient and Evolutionary Approaches to the Artificial Neural Networks Training According to Principles of Support Vector Machines

A gradient based learning for ANN training in pattern recognition tasks and a genetic approach for ANN pruning are proposed in this paper. The goal is to achieve a wide margin classifier the Vapnik-Chevornenkis (VC) dimension of which is being reduced in order to increase the generalization performance. Inspired by Support Vector Machines the examples closest to the decision boundary contribute to the training the most. The training penalty is rule-based and calculated according to the spatial distribution of the training examples relative to the separating hyperplane. The tendency to saturation of hidden neurons is suppressed. Genetic algorithm based method is proposed for reduction of the size of a trained ANN. The proposed algorithms were tested on artificial and real world data and compared to standard Backpropagation and Support Vector Machine with Gaussian RBF kernel.

The State of the Art in Computational Intelligence

Forewords: L.A. Zadeh: The Role of Fuzzy Systems in Computational Intelligence.- D.E. Goldberg: A Meditation on Computational Intelligence and Its Future.- K. Fukushima: Computational Intelligence and Modeling Neural Networks.- P. Sincak: Preface. Neural Networks: A. Chan, T. Spracklen: Discovering Common Features in Software Code Using Self-Organizing Maps.- N. Kopco, G.A. Carpenter: Graded Signal Functions for ARTMAP Neural Networks.- G. Andrejkova: Incremental Approximation by Layer Neural Networks.- V. Golovko, Y. Savitsky, T. Laopoulos, A. Sachenko, L. Grandinetti: Efficient Training of MLP with Training Step Rate Estimation.- I. Farkas: Self-Organizing Maps for Representing Structures.- R. Forgac, I. Mokris: Invariant Representation of Images by Pulse Coupled Neural Network.- S. Papadimitriou, L. Vladutu, S. Mavroudi, A. Bezerianos: Detection of Ischemic Episodes with a Combination of Unsupervised and Supervised Learning.- M. Jaszuk, W.A. Kaminski, A.D. Linkevich: Spatial Distribution of Patterns and the Hopfield Network Phase Space Geometry.- D. Hajtas, D. Durackova, G. Benyon-Tinker: Switched Capacitor-Based Integrate-And-Fire Neural Network.- J. Stefanovic: A Neural Network Algorithm for Digital Circuits Test Generation.- J. Ocenasek, J. Schwarz: The Parallel Bayesian Optimization Algorithm.- D. Krokavec, A. Filasova: Application of Heuristic Programming to Dynamic System Stabilization.- P. Geczy, S. Usui, J. Chmurny: First Order Dynamic Instance Selection. Fuzzy Systems: L.A. Zadeh: Toward an Enlargement of the Role of Natural Languages in Information Processing, Decision and Control.- M. Oussalah: Notes on Fusion of Uncertain Information.- K.-H. Temme, M. Fathi: Fix-Mundis for Fuzzy IF-THEN Rule Bases with T-Norm Based Compositional Rule of Inference Interpretation.- M. Navara, Z. Zabokrtsky: Computational Problems of Constrained Fuzzy Arithmetic.- S. Varga, M. Sabo: Linear Regression with Fuzzy Variables.- Z.M. Gacovski, G.M. Dimirovski, S. Deskovski: Fuzzy-Petri Net Reasoning System and Transfering of Knowledge to the Markov Chain.- T.H. Cao: Fuzzy Conceptual Graphs: A Language for Computational Intelligence Approaching Human Expression.- F. Botana, J. Ranilla, R. Mones, A. Bahamonde: A New Heuristic Measure for Learning Rules from Fuzzy Data.- B. De Baets: On the Role of Transitivity in Fuzzy Relational Calculus.- A. Slyeptsov, T. Tyshchuk: Project Network Planning on the Basis of Generalized Fuzzy Critical Path Method.- P. Marcincak, M. Matula: Flexible Querying - Data Structures Unification.- D. Smutna, P. Vojtas: New Connectives for (Full) Fuzzy Resolution.- I. Borgulya: Learn the Ranking of Precedence Cases.- A. Chodorek: Sugeno-Type Fuzzy Predictor of the MPEG Video Stream. Miscellaneous CI Techniques: F. Alexandre: Biological Inspiration for Multiple Memories Implementation and Cooperation.- J. Pospichal: Optimization as a Multisage Decision Making.- M. Dobes: Can Computers Be Conscious? Methodological Considerations from the Standpoint of the Humanities. Hybrid Systems: R. Matousek, P. Osmera, J. Roupec: GA-FIS for Dynamic Environment.- V. Olej, J. Krupka: Genetic Method for Optimization of Fuzzy Neural Networks.- Z. Raida: A Reverse Neural Model of a General Planar Transmission Line.- S. Girshgorn: Fuzzy Sets and the Theory of Neuronal Group Selection for the Problem of Interpretation.- D. Rutkowska, R. Nowicki, L. Rutkowski: Neuro-Fuzzy Architectures with Various Implication Operators.- P. Sincak, M. Hric, N. Kopco, J. Vascak: Fuzzy Cluster Identification Using Neural Networks. Applications and Case Studies: R. Blasko: Soft Computing Applications Developed by ECANSE.- M. Lukac, P. Bourgine: Tetris Player: Strategy Driven Algorithm.- P. Szathmary, M. Kolcun: Electrical Daily Load Forecasting Using Artificial Neural Network in the Power System of the Slovak Republic.- M. Hrehus, S. Figedy: Practical Approach to Prediction of Plant Technological

Enhancement of Classification Accuracy Using Conflation Procedures

Abstract This study focuses on conflation procedures to enhance classification accuracy for remote sensing imagery. In this context, conflation refers to the merging of the most accurate portions of a set of classified images to yield a result that is more accurate than any individual image. Two main types of conflation procedures are discussed. The first is a heuristic approach based on if‐then rules and the second is based on statistical manipulation of misclassification probabilities. The conflation procedures are tested on three classified images derived from a high‐resolution multispectral video mosaic using non‐traditional classification methods. For these test data, conflation yields improvements in classification accuracy of up to 15 percent.

Stochastic weight update in the backpropagation algorithm on feed-forward neural networks

We will examine stochastic weight update in the backpropagation algorithm on feed-forward neural networks. It was introduced by Salvetti and Wilamowski in 1994 in order to improve probability of convergence and speed of convergence. However, this update method has also one another quality, its implementation is simple for arbitrary network topology. In stochastic weight update scenario, constant number of weights is randomly selected and updated. This is in contrast to classical ordered update, where always all weights are updated. We will describe exact implementation, and present example results on toy-task data with feed-forward neural network topology. Stochastic weight update is suitable to replace classical ordered update without any penalty on implementation complexity and with good chance without penalty on quality of convergence.

Comparison Study of Robotic Middleware for Robotic Applications

Developing a robot system is a hard task and it requires a special programming knowledge. Moreover, the development and the operation of every robot system needs common tasks such as state management, communication among parts, synchronization, etc. Several software platforms were introduced for supporting these, hence the researcher and developer can concentrate the novel ideas. This document gives an overview of various robotic middleware. Strictly defined, robotic middleware serves for realizing the communication between various software components. In a wider sense, they are helping the process of development of robotic applications. The document contains descriptions of systems such as Robot Operating System (ROS), RT-Middleware, OPRoS and Orocos. The report also contains a comparison of the above-listed.

Design of LCL Filter Using Hybrid Intelligent Optimization for Photovoltaic System

This paper proposes new design method of LCL filter parameters by using hybrid intelligent optimization. Usually, the voltage-source inverters (VSI) have been used for control of both link voltage and power factor in the grid connected topologies. But, since VSI can cause high-order harmonics by high switching frequency, it is important to choose the filter parameters to achieve good filtering effect. Compared with traditional L and LC filter, LCL filter is known as an effective method on reducing harmonic distortion at the switching frequency of VSI. However, design of the LCL filter is complex, and the selection of the initial values of the inductors is difficult at the start of the design process. This paper proposes an approach for designing LCL filter parameters by hybrid optimization method with both genetic algorithm and clonal selection. Simulation results are provided to prove that the proposed method is more effective and simple than conventional methods.

Positional Fuzzy Relations in Robot Control

We define a new representation of the object in the known area. We can assign a name to the object and express uncertainty about its position using a fuzzy relation. This kind of representation about physical object can be used in the knowledge-based systems using rules and fuzzy inference system. We show an application in multi-robot coordination and control in the pursuit-evasion task.

Influence of Sci-Fi films on artificial intelligence and vice-versa

Sci-fi technological movie domain is an important part of human culture. The paper focus on comparison study of selected robotics sci-fi movie domain from a technological point of view. It is necessary to accomplish technological analysis of studied sci-fi movies and able to distinguish about possible current technology and future direction of the artificial intelligence in the domain of Robot intelligence. The review of existing movies which are in fact influencing thinking of humans is essential since it can influence a future research direction in AI. This information is interesting for inspiration of students and research associates in theory and applications. In conclusion, we envision potential problems of social networks and impact of Internet of things facts which is becoming a reality with IPv6 protocol. The goal of the paper is also to underline the importance of such cultural phenomena as sci-fi movies for the future of humanity.