Wyniki wyszukiwania - BazTech

1

Development of a neural statistical model for the prediction of relative humidity levels in the region of Rabat-Kenitra, North West Morocco

El Azhari Kaoutar, Abdallaoui Badreddine, Dehbi Ali, Abdalloui Abdelaziz, Zineddine Hamid

Journal of Water and Land Development

|

2022

|

no. 54

13--20

EN

This article accounts for the development of a powerful artificial neural network (ANN) model, designed for the prediction of relative humidity levels, using other meteorological parameters such as the maximum temperature, minimum temperature, precipitation, wind speed, and intensity of solar radiation in the Rabat-Kenitra region (a coastal area where relative humidity is a real concern). The model was applied to a database containing a daily history of five meteorological parameters collected by nine stations covering this region from 1979 to mid-2014. It has been demonstrated that the best performing three-layer (input, hidden, and output) ANN mathematical model for the prediction of relative humidity in this region is the multi-layer perceptron (MLP) model. This neural model using the Levenberg-Marquard algorithm, with an architecture of [5-11-1] and the transfer functions Tansig in the hidden layer and Purelin in the output layer, was able to estimate relative humidity values that were very close to those observed. This was affirmed by a low mean squared error (MSE) and a high correlation coefficient (R), compared to the statistical indicators relating to the other models developed as part of this study.

2

Automated Self-trained System of Functional Control and State Detection of Railway Transport Nodes

Akhmetov Bakhytzhan, Lakhno Valeriy, Oralbekova Ayaulym, Kaskatayev Zhanat, Mussayeva Gulmira

International Journal of Electronics and Telecommunications

|

2019

|

Vol. 65, No. 3

491--496

EN

Automation of data processing of contactless diagnostics (detection) of the technical condition of the majority of nodes and aggregates of railway transport (RWT) minimizes the damage from failures of these systems in operating modes. This becomes possible due to the rapid detection of serious defects at the stage of their origin. Basically, in practice, the control of the technical condition of the nodes and aggregates of the RWT is carried out during scheduled repairs. It is not always possible to identify incipient defects. Consequently, it is not always possible to warn personnel (machinists, repairmen, etc.) of significant damage to the RWT systems until their complete failure. The difficulties of obtaining diagnostic information is that there is interdependence between the main nodes of the RWT. This means that if physical damage occurs at any of the RWT nodes, in other nodes there can also occur malfunctions. As the main way to improve the efficiency of state detection of the nodes and aggregates of RWT, we see the direction of giving the adaptability property for an automated data processing system from various contactless diagnostic information removal systems. The global purpose can be achieved, in particular, through the use of machine learning methods and failure recognition (recognition objects). In order to improve the operational reliability and service life of the main nodes and aggregates of RWT, there are proposed an appropriate model and algorithm of machine learning of the operator control system of nodes and aggregates. It is proposed to use the Shannon normalized entropy measure and the Kullback-Leibler distance information criterion as a criterion of the learning effectiveness of the automated detection system and operator node state control of RWT. The article describes the application of the proposed method on the example of an automatic detection system (ADS) of the state of a traction motor of an electric locomotive. There are given the test data of the model and algorithm in the MATLAB environment.

3

Multidimensional neo-fuzzy-neuron for solving medical diagnostics tasks in online-mode

Mahmoud S. M. K., Perova I., Pliss I.

Journal of Applied Computer Science

|

2017

|

Vol. 25, nr 1

39--48

EN

In this paper neuro-fuzzy approach for medical data processing is considered. Special capacities for methods and systems of Computational Intelligence were introduced for Medical Data Mining tasks, like transparency and interpretability of obtained results, ability to classify nonconvex and overlapped classes that correspond to various diagnoses, necessity to process data in online mode and so on. Architecture based on the multidimensional neo-fuzzy-neuron was designed for situation of many diagnoses. For multidimensional neo-fuzzy-neuron adaptive learning algorithms that are a modification of Widrow-Hoff algorithm were introduced. This system was approbate on nervous system diseases data set from University of California Irvine (UCI) Repository and show high level of classification results.

4

Agent uczący się reguł świata Wumpusa

Beringer M., Kopyś M., Szymański J.

Zeszyty Naukowe Wydziału ETI Politechniki Gdańskiej. Technologie Informacyjne

|

2014

|

T. 22

43--57

PL

W artykule opisano implementację agenta uczącego się zasad panujących w świecie Wumpusa. W realizacji przyjęto założenie, że agent ma ograniczoną wiedzę, ktorą rozwija w miarę poruszania się po lochach. Przedstawione zostały rożne sposoby rozwiązania zagadnienia akwizycji wiedzy oraz podano szczegółowy opis zaimplementowanego algorytmu nauki. Przeprowadzono testy jakościowe wykonanej implementacji oraz dokonano dyskusji osiągniętych rezultatów.

EN

The article describes the implementation of an agent able to learn rules in Wumpus world. The implementation assumes that the agent has limited knowledge, which develops as it is navigating through the dungeons. Various solutions to knowledge acquisition related problems are presented and a detailed description of the implemented learning algorithm is provided. Qualitative tests of the implementation were made followed by the discussion of the achieved results.

5

Zastosowanie rozmytej mapy kognitywnej z dwukrokowym algorytmem uczenia opartym na markowskim modelu gradientu w predykcji szeregów czasowych

Jastriebov A., Poczęta K.

Logistyka

|

2014

|

nr 6

4905-4915

PL

Rozmyta mapa kognitywna FCM to inteligentny model, stanowiący efektywne narzędzie modelowania systemów predykcji szeregów czasowych. Zaletą FCM jest zdolność uczenia macierzy relacji na podstawie dostępnych danych z zastosowaniem nadzorowanych lub populacyjnych algorytmów. Niniejsza praca poświęcona jest analizie zastosowania dwukrokowego algorytmu uczenia rozmytej mapy kognitywnej bazującego na markowskim modelu gradientu w prognozowaniu szeregów czasowych. W procesie uczenia i testowania działania nauczonej FCM zastosowano dane idealne oraz dane z błędami pomiarowymi o różnym stopniu zaszumienia, wygenerowane na podstawie rzeczywistego systemu zarządzania biznesem elektronicznym. Dokonano analizy porównawczej dwukrokowej metody markowskiego modelu gradientu oraz jednokrokowej metody gradientowej pod kątem szybkości zbieżności algorytmu oraz uzyskanej dokładności predykcji szeregów czasowych.

EN

Fuzzy cognitive map FCM is an intelligent model, that can be used as a useful tool for modeling systems for time series prediction. The advantage of FCMs is their ability to learn the relations matrix based on real data with the use of supervised or population-based algorithms. This paper is devoted to the analysis of the application of FCM with two-step learning algorithm based on Markov model of gradient in time series prediction. Ideal data and data with measurement errors of varying degrees of noise were generated based on real-life system for e-commerce strategic planning and used in learning and testing process. The comparative analysis of two-step method of Markov model of gradient to one-step gradient method from the point of view of the speed of convergence of learning algorithm and the obtained precision of time series prediction was performed.

6

Analiza wybranych algorytmów uczenia w relacyjnej rozmytej mapie kognitywnej

Słoń G.

TTS Technika Transportu Szynowego

|

2012

|

R. 19, nr 9

1741--1751, CD

PL

Relacyjna rozmyta mapa kognitywna jest pewnym modelem matematycznym wykorzystującym metody inteligencji obliczeniowej, który może być stosowany do modelowania złożonych systemów charakteryzujących się nieprecyzyjnością. Kluczowym problemem w tego rodzaju zagadnieniach jest adaptacja parametrów (uczenie) modelu w sposób, dzięki któremu jego praca będzie jak najdokładniej odwzorowywać zachowanie rzeczywistego obiektu. W artykule przedstawiono podstawowe informacje dotyczące problematyki uczenia modelu relacyjnej rozmytej mapy kognitywnej oraz efekty stosowania różnych podejść do uczenia w zależności od celów modelowania.

EN

Relational fuzzy cognitive map is a certain mathematical model which uses some methods of the computational intelligence and can be used to model complex systems characterized by imprecision. A key problem in this kind of issues is an adaptation of parameters of the model (learning) in the way in which his work will imitate as closely as possible the behavior of a real object. The article presents basic information about problems of the model of fuzzy relational cognitive map learning and the effects of different approaches to the learning in dependence on the modeling purposes.

7

Model-building adaptive critics for semi-Markov control

Gosavi A., Murray S., Hu J., Ghosh S.

Journal of Artificial Intelligence and Soft Computing Research

|

2012

|

Vol. 2, No. 1

43--58

EN

Adaptive (or actor) critics are a class of reinforcement learning algorithms. Generally, in adaptive critics, one starts with randomized policies and gradually updates the probability of selecting actions until a deterministic policy is obtained. Classically, these algorithms have been studied for Markov decision processes under model-free updates. Algorithms that build the model are often more stable and require less training in comparison to their model-free counterparts. We propose a new model-building adaptive critic, which builds the model during the learning, for a discounted-reward semi-Markov decision process under some assumptions on the structure of the process. We illustrate the use of our algorithm with numerical results on a system with 10 states and a real-world case-study from management science.

8

Learning Workflow Petri Nets

Esparza J., Leucker M., Schlund M.

Fundamenta Informaticae

|

2011

|

Vol. 113, nr 3/4

205-228

EN

Workflow mining is the task of automatically producing a workflow model from a set of event logs recording sequences of workflow events; each sequence corresponds to a use case or workflow instance. Formal approaches to workflow mining assume that the event log is complete (contains enough information to infer the workflow) which is often not the case. We present a learning approach that relaxes this assumption: if the event log is incomplete, our learning algorithm automatically derives queries about the executability of some event sequences. If a teacher answers these queries, the algorithm is guaranteed to terminate with a correct model. We provide matching upper and lower bounds on the number of queries required by the algorithm, and report on the application of an implementation to some examples.

9

Sztuczne sieci neuronowe - podstawowe struktury sieciowe i algorytmy uczące

Osowski S.

Przegląd Elektrotechniczny

|

2009

|

R. 85, nr 8

1-8

PL

Praca przedstawia struktury i algorytmy uczenia trzech podstawowych rodzajów sieci neuronowych: sieci sigmoidalnej wielowarstwowej MLP, sieci RBF oraz SVM. Sieci te pełnią podobną rolę uniwersalnego aproksymatora zmiennych wielowymiarowych, różniąc się przede wszystkim rodzajem zastosowanych neuronów i algorytmem uczącym. Pokazano uniwersalność tych rozwiązań i ich użyteczność w wielu problemach praktycznych występujących w technice.

EN

The paper presents the overview of the basic structures and learning algorithms of the most typical supervised neural networks. To the most important belong multilayer perceptron, radial basis function network and support vector machine. All of them fulfill the role of the universal approximators. The paper shows the universal character of these solutions and their applicability in solving different problems of various characters.

10

Global optimization based learning for recurrent neural networks

Strzelczyk M., Domański P. D.

Prace Naukowe Politechniki Warszawskiej. Elektronika

|

2002

|

z. 139

163--172

EN

This chapter describes the application of two different global optimization algorithms (CRS3 and the evolutionary strategy) to the problem of recurrent neural network learning. The performance has been compared with the standard multilayer perceptron with classical gradient learning strategy. The performance has been tested on different dynamic test models. The efficiency of the global optimization approach is shown.

11

Learning algorithms for a class of knowledge-based systems with dynamical knowledge representation.

Bubnicki Z.

Systems Science

|

2000

|

Vol. 26, nr 1

15-27

EN

The paper deals with a specific class of knowledge-based systems with a dynamical plant described by a knowledge representation with unknown parameters. The learning consists of step by step knowledge evaluation and updating, and using the results for the determination of the current control decisions. The paper presents an extension of methods and algorithms described by the author (1999) for the knowledge-based learning systems with a static plant. Two forms of the plant descriptions are taken into account: the relational knowledge representation, and the logical knowledge representation. A simple illustrative example is given.