Wyniki wyszukiwania - BazTech

1

Digraph-building method for finding a set of minimal realizations of 2-D dynamic systems

Markowski K. A., Hryniów K.

Bulletin of the Polish Academy of Sciences. Technical Sciences

|

2018

|

Vol. 66, nr 5

589--597

EN

This paper presents a digraph-building method designed to find the determination of realization of two-dimensional dynamic system. The main differences between the method proposed and other state-of-the-art solutions used include finding a set of realizations (belonging to a defined class) instead of only one realization, and the fact that obtained realizations have minimal size of state matrices. In the article, the proposed method is described, compared to state-of-the-art methods and illustrated with numerical examples. To the best of authors’ knowledge, the method shown in the paper is superior to all other state-of-the-art solutions both in terms of number of solutions and their matrix size. Additionally, MATLAB function for determination of realization based on the set of state matrices is included.

2

Parallel Digraphs-building Computer Algorithm for Finding a Set of Characteristic Polynomial Realisations of Dynamic System

Hryniów K., Markowski K. A.

Journal of Automation Mobile Robotics and Intelligent Systems

|

2016

|

Vol. 10, No. 3

38--51

EN

This paper presents in-depth the parallel computer algorithm for the determination of characteristic polynomial realisations of dynamic system. The main differences between the depicted method and other state of- the-art solutions include finding not few realisations, but a whole set, and the fact that the found realisations are always minimal among all possible. As digraphsbuilding methods used in the algorithm are NP-complete or NP-hard problems, the algorithm is paralleled and GPGPU (General-Purpose computing on Graphics Processor Units) computation is proposed as the only feasible solution. The article describes in detail the proposed method, discusses it’s complexity, presents optimisation solutions and still open problems. The working algorithm is illustrated with a numerical example and compared to results of other known methods.

3

Probabilistic Sequence Mining : Evaluation and Extension of ProMFS Algorithm for Real-Time Problems

Hryniów K., Dzieliński A.

International Journal of Electronics and Telecommunications

|

2012

|

Vol. 58, No. 4

323-326

EN

Sequential pattern mining is an extensively studied method for data mining. One of new and less documented approaches is estimation of statistical characteristics of sequence for creating model sequences, that can be used to speed up the process of sequence mining. This paper proposes extensive modifications to one of such algorithms, ProMFS (probabilistic algorithm for mining frequent sequences), which notably increases algorithm's processing speed by a significant reduction of its computational complexity. A new version of algorithm is evaluated for real-life and artificial data sets and proven to be useful in real-time applications and problems.

4

Negative feature selection algorithm for anomaly detection in real time

Hryniów K., Dzieliński A.

Studia Informatica : systems and information technology

|

2011

|

Vol. 1-2(15)

15--23

EN

Anomaly detection methods are of common use in many fields, including databases and large computer systems. This article presents new algorithm based on negative feature selection, which can be used to find anomalies in real time. Proposed algorithm, called Negative Feature Selection algorithm (NegFS) can be also used as first step for preprocessing data analyzed by neural networks, rule-based systems or other anomaly detection tools, to speed up the process for large and very large datasets of different types.