Tomáš Novosád

Vector model improvement using suffix trees

There are many ways how to search for documents in document collections. These methods take advantage of Boolean, vector, probabilistic and other models for representation of documents, queries, rules and procedures which can determine correspondence between user requests and documents. Each of these models have several restrictions. These restrictions do not allow a user to find all relevant documents. There are many irrelevant documents among returned ones by the system and some relevant documents missing at all. In the article there is a new method suggested which uses suffix trees for the vector query improvement. This method treats with documents as a, set of phrases (sentences) not just as a set of words. The sentence has a specific, semantic meaning (words in the sentence are ordered). This is advantage in comparison with the treated document just like with, a bag of words.

Artificially evolved soft computing models for photovoltaic power plant output estimation

Renewable energy sources are becoming a significant part of todays energy mix. The unstable production of many renewable energy sources including photovoltaic and wind power plants puts increased demands on power transmission systems and on the power grid as a whole. Soft computing methods can contribute to the prediction of electric energy production of renewable resources and therefore to the reliability of the power transmission networks. This work compares two soft computing methods that utilize genetic programming to evolve predictors of a selected renewable energy resource that meets the real world criterion of high output variance and relatively large installed power (in context of the power distribution system of the Czech Republic).

Prosima: Protein similarity algorithm

In this article, we present a novel algorithm for measuring protein similarity based on their three dimensional structure (protein tertiary structure). The PROSIMA algorithm using suffix tress for discovering common parts of main-chains of all proteins appearing in current NCSB protein data bank (PDB). By identifying these common parts we build a vector model and next use classical information retrieval tasks based on the vector model to measure the similarity between proteins - all to all protein similarity. For the calculation of protein similarity we are using tf-idf term weighing schema and cosine similarity measure. The goal of this work to use the whole current PDB database (downloaded on June 2009) of known proteins, not just some kinds of selections of this database, which have been studied in other works. We have chose the SCOP database for verification of precision of our algorithm because it is maintained primarily by humans. The next success of this work is to be able to determine protein SCOP categories of proteins not included in the latest version of the SCOP database (v. 1.75) with nearly 100% precision.

Photovoltaic power plant output estimation by neural networks and fuzzy inference

The stochastic production of renewable energy sources puts increased demands on power grids worldwide. Neurocomputing methods can be used for the forecast of electric energy production of renewable resources and contribute to the reliability of energy systems. This study compares two neurocomputing methods as predictors of a selected photovoltaic power plant in the Czech Republic that meets the real world criterion of high output variance and relatively large installed power.

Mobile Phone Positioning in GSM Networks Based on Information Retrieval Methods and Data Structures

In this article we present a novel method for mobile phone positioning using a vector space model, suffix trees and an information retrieval approach. The method works with parameters which can be acquired from any common mobile phone without the necessity of installing additional hardware and is handset based. The algorithm is based on a database of previous measurements which are used as an index which looks for the nearest neighbor toward the query measurement. The accuracy of the algorithm is in most cases good enough to accomplish the E9-1-1 requirements on tested data.

Fast intrusion detection system based on Flexible Neural Tree

Computer security is very important in these days. Computers are used probably in any industry and their protection against attacks is very important task. The protection usually consist in several levels. The first level is preventions. Intrusion detection system (IDS) may be used as next level. IDS is useful in detection of intrusions, but also in monitoring of security issues and the traffic. This paper present IDS based on Flexible Neural Trees. Flexible neural tree is hierarchical neural network, which is automatically created using evolutionary algorithms to solving of defined problem. This is very important, because it is not necessary to set the structure and the weights of neural networks prior the problem is solved. The accuracy of proposed technique is always above 98% and the speed of decision making process enable its using in real-time applications.

Heavy facilities tension prediction using Flexible Neural Trees

In this article we show the usage of soft-computing methods to solve the real problem of computation of tension in facilities working under very hard conditions in industrial environment. Because the classical mathematical approaches such as Finite Element Method (FEM) are very time consuming, the more progressive soft-computing methods are on the place. We have proposed two step algorithm based on Flexible Neural Tree (FNT) and Particle Swarm Optimization (PSO) which is more efficient then typical approach (FEM). Flexible neural tree is hierarchical neural network like structure, which is automatically created and optimized using evolutionary like algorithms to solve the given problem. This is very important, because it is not necessary to set the structure and the weights of neural networks prior the problem is solved. The accuracy of proposed technique is good enough to be used in real environments.

Evaluation of Novel Soft Computing Methods for the Prediction of the Dental Milling Time-Error Parameter

This multidisciplinary study presents the application of two well known soft computing methods – flexible neural trees, and evolutionary fuzzy rules – for the prediction of the error parameter between real dental milling time and forecast given by the dental milling machine. In this study a real data set obtained by a dynamic machining center with five axes simultaneously is analyzed to empirically test the novel system in order to optimize the time error.

Prediction of dental milling time-error by flexible neural trees and fuzzy rules

This multidisciplinary study presents the application of two soft computing methods utilizing the artificial evolution of symbolic structures --- evolutionary fuzzy rules and flexible neural trees --- for the prediction of dental milling time-error, i.e. the error between real dental milling time and forecast given by the dental milling machine. In this study a real data set obtained by a dynamic machining center with five axes simultaneously is analyzed to empirically test the novel system in order to optimize the time error.

CLUSTERING THE MOBILE PHONE POSITIONS BASED ON SUFFIX TREE AND SELF-ORGANIZING MAPS

In this article we present a novel method for mobile phone positioning using a vector space model, suffix trees and an information retrieval approach. The algorithm is based on a database of previous measurements which are used as an index which looks for the nearest neighbor toward the query measurement. The accuracy of the algorithm is, in most cases, good enough to accomplish the E9-1-1 standards requirements on tested data. In addition, we are trying to look at the clusters of patterns that we have created from measured data and we have reflected them to the map. We use Self-Organizing Maps for these purposes.