Informacja

Drogi użytkowniku, aplikacja do prawidłowego działania wymaga obsługi JavaScript. Proszę włącz obsługę JavaScript w Twojej przeglądarce.

Wyszukujesz frazę "recurrent neural network" wg kryterium: Temat


Tytuł:
Heuristic modeling of objects and processes using dynamic neural networks
Heurystyczne modelowanie obiektów i procesów przy pomocy dynamicznych sieci neuronowych
Autorzy:
Przystałka, P.
Powiązania:
https://bibliotekanauki.pl/articles/327816.pdf
Data publikacji:
2006
Wydawca:
Polska Akademia Nauk. Polskie Towarzystwo Diagnostyki Technicznej PAN
Tematy:
sztuczna sieć neuronowa
lokalnie rekurencyjna sieć neuronowa
systemy dynamiczne
metoda quasi-Newtonowska
modelowanie heurystyczne
artificial neural network
locally recurrent neural network
dynamic systems
quasi-Newton methods
heuristic modelling
Opis:
The methodology of heuristic modeling is one of the subjects included in the activities developed by the Department of Fundamentals of Machinery Design [4, 6]. Among all the approaches of heuristic modeling some of the most common are artificial neural networks. There are many papers and books devoted to applications of neural networks for modeling dynamic systems [1, 2, 4, 5, 6, 7]. In this paper, known approach basing on dynamic neuron model is presented (dynamic neuron with IIR filter in the activation block [2]) but some developments are introduced. Locally recurrent networks which are composed of dynamic neural units described in [2, 5, 7] are able to model behavior of complex dynamic systems. Nevertheless, they have one major disadvantage, that is, neural networks composed of these neurons are not able to represent stochastic behaviors of some objects [4,6]. By introducing the ARMAX (or ARX) system into dynamic neuron model author has received dynamic neuron unit that never behaves in the same way (it brings an artificial neuron closer and closer to the biological model). In this paper the author presents formal description of dynamic neuron unit with ARMAX system in the feedback block. There are also described a general structure of dynamic neural network composed of these neurons, two known training methods and some commonly used quality measures. At the end of the paper three examples of applications are given.
Metodologia heurystycznego modelowania obiektów i procesów jest jednym z kierunków badań rozwijanym prze Katedrę Podstaw Konstrukcji Maszyn [4, 6]. Spośród wielu metod modelowania heurystycznego duże znaczenie odgrywają metody bazujące na sztucznych sieciach neuronowych. Można wyróżnić wiele ciekawych prac badawczych prowadzonych w kierunku modelowania systemów dynamicznych z zastosowaniem tego typu narzędzia [1, 2, 4, 5, 6, 7]. W artykule zaprezentowano znane podejście bazujące na dynamicznych neuronach (dynamiczny neuron z filtrem IIR w bloku aktywacyjnym [2]) z pewnymi modyfikacjami. Lokalnie rekurencyjne sieci neuronowe złożone z dynamicznych neuronów opisane w [2, 5, 7] nadają się do modelowania zachowania złożonych systemów dynamicznych. Jednakże, posiadają one jedną główną wadę tzn. nie są zdolne do reprezentowania zachowania losowego niektórych obiektów [4, 6]. Poprzez wprowadzenie systemu typu ARMAX (ARX) do modeli dynamicznych neuronów autor otrzymał dynamiczny model neuronu, który nigdy nie zachowują się w ten sam sposób (przybliża to model sztucznego neuronu do jego biologicznego wzoru). W artykule autor prezentuje formalny opis dynamicznego neuronu z systemem typu ARMAX w bloku sprzężenie zwrotnego. Opisuje również ogólną strukturę dynamicznej sieci neuronowej złożonej z tych neuronów, dwa znane algorytmy trenujące oraz powszechnie stosowane miary jakości. Przykładowe zastosowania opisywanych sieci zaprezentowane są w końcowym fragmencie opracowania.
Źródło:
Diagnostyka; 2006, 2(38); 31-36
1641-6414
2449-5220
Pojawia się w:
Diagnostyka
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Recurrent neural networks for dynamic reliability analysis
Autorzy:
Cadini, F.
Zio, E.
Pedroni, N.
Powiązania:
https://bibliotekanauki.pl/articles/2069583.pdf
Data publikacji:
2007
Wydawca:
Uniwersytet Morski w Gdyni. Polskie Towarzystwo Bezpieczeństwa i Niezawodności
Tematy:
dynamic reliability analysis
infinite impulse response
locally recurrent neural network
long-term non-linear dynamics
system state memory
simplified nuclear reactor
Opis:
A dynamic approach to the reliability analysis of realistic systems is likely to increase the computational burden, due to the need of integrating the dynamics with the system stochastic evolution. Hence, fast-running models of process evolution are sought. In this respect, empirical modelling is becoming a popular approach to system dynamics simulation since it allows identifying the underlying dynamic model by fitting system operational data through a procedure often referred to as ‘learning’. In this paper, a Locally Recurrent Neural Network (LRNN) trained according to a Recursive Back-Propagation (RBP) algorithm is investigated as an efficient tool for fast dynamic simulation. An application is performed with respect to the simulation of the non-linear dynamics of a nuclear reactor, as described by a simplified model of literature.
Źródło:
Journal of Polish Safety and Reliability Association; 2007, 1; 45--53
2084-5316
Pojawia się w:
Journal of Polish Safety and Reliability Association
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Local stability conditions for discrete-time cascade locally recurrent neural networks
Autorzy:
Patan, K.
Powiązania:
https://bibliotekanauki.pl/articles/907772.pdf
Data publikacji:
2010
Wydawca:
Uniwersytet Zielonogórski. Oficyna Wydawnicza
Tematy:
sieć lokalnie rekurencyjna
stabilność
stabilizacja
uczenie się
optymalizacja ograniczona
locally recurrent neural network
stability
stabilization
learning
constrained optimization
Opis:
The paper deals with a specific kind of discrete-time recurrent neural network designed with dynamic neuron models. Dynamics are reproduced within each single neuron, hence the network considered is a locally recurrent globally feedforward. A crucial problem with neural networks of the dynamic type is stability as well as stabilization in learning problems. The paper formulates local stability conditions for the analysed class of neural networks using Lyapunov's first method. Moreover, a stabilization problem is defined and solved as a constrained optimization task. In order to tackle this problem, a gradient projection method is adopted. The efficiency and usefulness of the proposed approach are justified by using a number of experiments.
Źródło:
International Journal of Applied Mathematics and Computer Science; 2010, 20, 1; 23-34
1641-876X
2083-8492
Pojawia się w:
International Journal of Applied Mathematics and Computer Science
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Recurrent neural identification and control of a continuous bioprocess via first and second order learning
Autorzy:
Baruch, I.
Mariaca-Gaspar, C. R.
Powiązania:
https://bibliotekanauki.pl/articles/385133.pdf
Data publikacji:
2010
Wydawca:
Sieć Badawcza Łukasiewicz - Przemysłowy Instytut Automatyki i Pomiarów
Tematy:
backpropagation learning
direct adaptive neural control
indirect adaptive sliding mode control
Kalman filter recurrent neural network identifier
Levenberg-Marquardt learning
Opis:
This paper applies a new Kalman Filter Recurrent Neural Network (KFRNN) topology and a recursive Levenberg-Mar quardt (L-M) learning algorithm capable to estimate para meters and states of highly nonlinear unknown plant in noisy environment. The proposed KFRNN identifier, learned by the Backpropagation and L-M learning algorithm, was incorporated in a direct and indirect adaptive neural con trol schemes. The proposed control schemes were applied for real-time recurrent neural identification and control of a continuous stirred tank bioreactor model, where fast convergence, noise filtering and low mean squared error of reference tracking were achieved.
Źródło:
Journal of Automation Mobile Robotics and Intelligent Systems; 2010, 4, 4; 37-52
1897-8649
2080-2145
Pojawia się w:
Journal of Automation Mobile Robotics and Intelligent Systems
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Robustifying analysis of the direct adaptive control of unknown multivariable nonlinear systems based on a new neuro-fuzzy method
Autorzy:
Theodoridis, D. C.
Boutalis, Y.S.
Christodoulou, M. A.
Powiązania:
https://bibliotekanauki.pl/articles/91598.pdf
Data publikacji:
2011
Wydawca:
Społeczna Akademia Nauk w Łodzi. Polskie Towarzystwo Sieci Neuronowych
Tematy:
nonlinear systems
control
neuro-fuzzy dynamical system
fuzzy systems
FS
fuzzy recurrent high order neural network
F-RHONN
adaptive regulator
parameter
“Hopping”
“Modified Hopping”
modeling errors
asymptotic regulation
Opis:
In this paper, we are dealing with the problem of directly regulating unknown multivariable affine in the control nonlinear systems and its robustness analysis. The method employs a new Neuro-Fuzzy Dynamical System definition, which uses the concept of Fuzzy Systems (FS) operating in conjunction with High Order Neural Networks. In this way the unknown plant is modeled by a fuzzy - recurrent high order neural network structure (F-RHONN), which is of the known structure considering the neglected nonlinearities. The development is combined with a sensitivity analysis of the closed loop in the presence of modeling imperfections and provides a comprehensive and rigorous analysis showing that our adaptive regulator can guarantee the convergence of states to zero or at least uniform ultimate boundedness of all signals in the closed loop when a not-necessarily-known modeling error is applied. The existence and boundedness of the control signal is always assured by employing a method of parameter “Hopping” and “Modified Hopping”, which appears in the weight updating laws. Simulations illustrate the potency of the method showing that by following the proposed procedure one can obtain asymptotic regulation despite the presence of modeling errors. Comparisons are also made to simple recurrent high order neural network (RHONN) controllers, showing that our approach is superior to the case of simple RHONN’s.
Źródło:
Journal of Artificial Intelligence and Soft Computing Research; 2011, 1, 1; 59-79
2083-2567
2449-6499
Pojawia się w:
Journal of Artificial Intelligence and Soft Computing Research
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
A class of neuro-computational methods for assamese fricative classification
Autorzy:
Patgiri, C.
Sarma, M.
Sarma, K. K.
Powiązania:
https://bibliotekanauki.pl/articles/91763.pdf
Data publikacji:
2015
Wydawca:
Społeczna Akademia Nauk w Łodzi. Polskie Towarzystwo Sieci Neuronowych
Tematy:
neuro-computational classifier
fricative phonemes
Assamese language
Recurrent Neural Network
RNN
neuro fuzzy classifier
linear prediction cepstral coefficients
LPCC
self-organizing map
SOM
adaptive neuro-fuzzy inference system
ANFIS
klasyfikator neuronowy
klasyfikator neuronowo rozmyty
sieć Kohonena
Opis:
In this work, a class of neuro-computational classifiers are used for classification of fricative phonemes of Assamese language. Initially, a Recurrent Neural Network (RNN) based classifier is used for classification. Later, another neuro fuzzy classifier is used for classification. We have used two different feature sets for the work, one using the specific acoustic-phonetic characteristics and another temporal attributes using linear prediction cepstral coefficients (LPCC) and a Self Organizing Map (SOM). Here, we present the experimental details and performance difference obtained by replacing the RNN based classifier with an adaptive neuro fuzzy inference system (ANFIS) based block for both the feature sets to recognize Assamese fricative sounds.
Źródło:
Journal of Artificial Intelligence and Soft Computing Research; 2015, 5, 1; 59-70
2083-2567
2449-6499
Pojawia się w:
Journal of Artificial Intelligence and Soft Computing Research
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Dynamic network functional comparison via approximate-bisimulation
Autorzy:
Donnarumma, F.
Murano, A.
Prevete, R.
Powiązania:
https://bibliotekanauki.pl/articles/206758.pdf
Data publikacji:
2015
Wydawca:
Polska Akademia Nauk. Instytut Badań Systemowych PAN
Tematy:
continuous time recurrent neural network
dynamic networks
bisimulation
network equivalence
Opis:
It is generally unknown how to formally determine whether different neural networks have a similar behaviour. This question intimately relates to the problem of finding a suitable similarity measure to identify bounds on the input-output response distances of neural networks, which has several interesting theoretical and computational implications. For example, it can allow one to speed up the learning processes by restricting the network parameter space, or to test the robustness of a network with respect to parameter variation. In this paper we develop a procedure that allows for comparing neural structures among them. In particular, we consider dynamic networks composed of neural units, characterised by non-linear differential equations, described in terms of autonomous continuous dynamic systems. The comparison is established by importing and adapting from the formal verification setting the concept of δ−approximate bisimulations techniques for non-linear systems. We have positively tested the proposed approach over continuous time recurrent neural networks (CTRNNs).
Źródło:
Control and Cybernetics; 2015, 44, 1; 99-127
0324-8569
Pojawia się w:
Control and Cybernetics
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Basic concepts of dynamic recurrent neural networks development
Autorzy:
Boyko, N.
Pobereyko, P.
Powiązania:
https://bibliotekanauki.pl/articles/410971.pdf
Data publikacji:
2016
Wydawca:
Polska Akademia Nauk. Oddział w Lublinie PAN
Tematy:
recurrent neural network
dynamic system
learning algorithms
reservoir computing
unsteady dynamics
Opis:
In this work formulated relevance, set out an analytical review of existing approaches to the research recurrent neural networks (RNN) and defined precondition appearance a new direction in the field neuroinformatics – reservoir computing. Shows generalized classification neural network (NN) and briefly described main types dynamics and modes RNN. Described topology, structure and features of the model NN with different nonlinear functions and with possible areas of progress. Characterized and systematized wellknown learning methods RNN and conducted their classification by categories. Determined the place RNN with unsteady dynamics of other classes RNN. Deals with the main parameters and terminology, which used to describe models RNN. Briefly described practical implementation recurrent neural networks in different areas natural sciences and humanities, and outlines and systematized main deficiencies and the advantages of using different RNN. The systematization of known recurrent neural networks and methods of their study is performed and on this basis the generalized classification of neural networks was proposed.
Źródło:
ECONTECHMOD : An International Quarterly Journal on Economics of Technology and Modelling Processes; 2016, 5, 2; 63-68
2084-5715
Pojawia się w:
ECONTECHMOD : An International Quarterly Journal on Economics of Technology and Modelling Processes
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
An arma type pi-sigma artificial neural network for nonlinear time series forecasting
Autorzy:
Akdeniz, E.
Egrioglu, E.
Bas, E.
Yolcu, U.
Powiązania:
https://bibliotekanauki.pl/articles/91816.pdf
Data publikacji:
2018
Wydawca:
Społeczna Akademia Nauk w Łodzi. Polskie Towarzystwo Sieci Neuronowych
Tematy:
high order artificial neural networks
pi-sigma neural network, forecasting
recurrent neural network
particle swarm optimization (PSO)
Opis:
Real-life time series have complex and non-linear structures. Artificial Neural Networks have been frequently used in the literature to analyze non-linear time series. High order artificial neural networks, in view of other artificial neural network types, are more adaptable to the data because of their expandable model order. In this paper, a new recurrent architecture for Pi-Sigma artificial neural networks is proposed. A learning algorithm based on particle swarm optimization is also used as a tool for the training of the proposed neural network. The proposed new high order artificial neural network is applied to three real life time series data and also a simulation study is performed for Istanbul Stock Exchange data set.
Źródło:
Journal of Artificial Intelligence and Soft Computing Research; 2018, 8, 2; 121-132
2083-2567
2449-6499
Pojawia się w:
Journal of Artificial Intelligence and Soft Computing Research
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Forecasting economic and financial indicators by supply of deep and recovery neural networks
Autorzy:
Boyko, N.
Ivanets, A.
Bosik, M.
Powiązania:
https://bibliotekanauki.pl/articles/411261.pdf
Data publikacji:
2018
Wydawca:
Polska Akademia Nauk. Oddział w Lublinie PAN
Tematy:
neural network
deep
recurrent
activation function
feedforward
neuron
hidden layer
stock price prediction
Opis:
This paper studies the potential of the application of the Recurrent Neural Networks, as well as the Deep Neural Networks in the field of the finances and trading. In particular, their use in the stock price predicting software. The concepts of the RNNs and DNNs are provided and explained thoroughly. Both techniques RNNs and DNNs are utilized in the implementation of the stock price predicting software. Two separate versions of the software are created in order to demonstrate the main differences between the algorithms, as well as to determine the best of the two. Each version is thoroughly examined. The comparison of each of the algorithms is performed and highlighted. Examples of the implementations of the software, utilizing each of the algorithms on big volumes of stock data, for stock price prediction are provided. The article summarizes the concept of stock price prediction backed by the popular machine learning algorithms and its application in the nowadays world.
Źródło:
ECONTECHMOD : An International Quarterly Journal on Economics of Technology and Modelling Processes; 2018, 7, 2; 3-8
2084-5715
Pojawia się w:
ECONTECHMOD : An International Quarterly Journal on Economics of Technology and Modelling Processes
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Character-based recurrent neural networks for morphological relational reasoning
Autorzy:
Mogren, Olof
Johansson, Richard
Powiązania:
https://bibliotekanauki.pl/articles/103847.pdf
Data publikacji:
2019
Wydawca:
Polska Akademia Nauk. Instytut Podstaw Informatyki PAN
Tematy:
morphological analogies
morphological inflection
morphological reinflection
recurrent neural network
character-based modelling
Opis:
We present a model for predicting inflected word forms based on morphological analogies. Previous work includes rule-based algorithms that determine and copy affixes from one word to another, with limited support for varying inflectional patterns. In related tasks such as morphological reinflection, the algorithm is provided with an explicit enumeration of morphological features which may not be available in all cases. In contrast, our model is feature-free: instead of explicitly representing morphological features, the model is given a demo pair that implicitly specifies a morphological relation (such as write: writes specifying infinitive:present). Given this demo relation and a query word (e.g. watch), the model predicts the target word (e.g. watches). To address this task, we devise a character-based recurrent neural network architecture using three separate encoders and one decoder. Our experimental evaluation on five different languages shows that the exact form can be predicted with high accuracy, consistently beating the baseline methods. Particularly, for English the prediction accuracy is 94.85%. The solution is not limited to copying affixes from the demo relation, but generalizes to words with varying inflectional patterns, and can abstract away from the orthographic level to the level of morphological forms.
Źródło:
Journal of Language Modelling; 2019, 7, 1; 139-170
2299-856X
2299-8470
Pojawia się w:
Journal of Language Modelling
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Comparison of optimization algorithms of connectionist temporal classifier for speech recognition system
Porównanie algorytmów optymalizacji klasyfikatora czasowego do systemu rozpoznawania mowy
Autorzy:
Amirgaliyev, Yedilkhan
Darkhan, Kuanyshbay
Shoiynbek, Aisultan
Powiązania:
https://bibliotekanauki.pl/articles/408796.pdf
Data publikacji:
2019
Wydawca:
Politechnika Lubelska. Wydawnictwo Politechniki Lubelskiej
Tematy:
recurrent neural network
search method
acoustic
systems modeling language
rekurencyjna sieć neuronowa
metoda wyszukiwania
akustyka
język modelowania systemów
Opis:
This paper evaluates and compares the performances of three well-known optimization algorithms (Adagrad, Adam, Momentum) for faster training the neural network of CTC algorithm for speech recognition. For CTC algorithms recurrent neural network has been used, specifically Long- Short-Term memory. LSTM is effective and often used model. Data has been downloaded from VCTK corpus of Edinburgh University. The results of optimization algorithms have been evaluated by the Label error rate and CTC loss.
W artykule dokonano oceny i porównania wydajności trzech znanych algorytmów optymalizacyjnych (Adagrad, Adam, Momentum) w celu przyspieszenia treningu sieci neuronowej algorytmu CTC do rozpoznawania mowy. Dla algorytmów CTC wykorzystano rekurencyjną sieć neuronową, w szczególności LSTM, która jest efektywnym i często używanym modelem. Dane zostały pobrane z wydziału VCTK Uniwersytetu w Edynburgu. Wyniki algorytmów optymalizacyjnych zostały ocenione na podstawie wskaźników Label error i CTC loss.
Źródło:
Informatyka, Automatyka, Pomiary w Gospodarce i Ochronie Środowiska; 2019, 9, 3; 54-57
2083-0157
2391-6761
Pojawia się w:
Informatyka, Automatyka, Pomiary w Gospodarce i Ochronie Środowiska
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Detecting objects using Rolling Convolution and Recurrent Neural Network
Autorzy:
Huang, WenQing
Huang, MingZhu
Wang, YaMing
Powiązania:
https://bibliotekanauki.pl/articles/226942.pdf
Data publikacji:
2019
Wydawca:
Polska Akademia Nauk. Czytelnia Czasopism PAN
Tematy:
multi-scale features
global context information
rolling convolution
recurrent neural network
Opis:
At present, most of the existing target detection algorithms use the method of region proposal to search for the target in the image. The most effective regional proposal method usually requires thousands of target prediction areas to achieve high recall rate.This lowers the detection efficiency. Even though recent region proposal network approach have yielded good results by using hundreds of proposals, it still faces the challenge when applied to small objects and precise locations. This is mainly because these approaches use coarse feature. Therefore, we propose a new method for extracting more efficient global features and multi-scale features to provide target detection performance. Given that feature maps under continuous convolution lose the resolution required to detect small objects when obtaining deeper semantic information; hence, we use rolling convolution (RC) to maintain the high resolution of low-level feature maps to explore objects in greater detail, even if there is no structure dedicated to combining the features of multiple convolutional layers. Furthermore, we use a recurrent neural network of multiple gated recurrent units (GRUs) at the top of the convolutional layer to highlight useful global context locations for assisting in the detection of objects. Through experiments in the benchmark data set, our proposed method achieved 78.2% mAP in PASCAL VOC 2007 and 72.3% mAP in PASCAL VOC 2012 dataset. It has been verified through many experiments that this method has reached a more advanced level of detection.
Źródło:
International Journal of Electronics and Telecommunications; 2019, 65, 2; 293-301
2300-1933
Pojawia się w:
International Journal of Electronics and Telecommunications
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
A Study of Correlation between Fishing Activity and AIS Data by Deep Learning
Autorzy:
Shen, K. Y.
Chu, Y. J.
Chang, S. J.
Chang, S. M.
Powiązania:
https://bibliotekanauki.pl/articles/1841621.pdf
Data publikacji:
2020
Wydawca:
Uniwersytet Morski w Gdyni. Wydział Nawigacyjny
Tematy:
AIS Data
deep learning framework
learning methods
Recurrent Neural Network
(RNN)
Automatic Identification System
(AIS)
fishing operation
Opis:
Previous researches on the prediction of fishing activities mainly rely on the speed over ground (SOG) as the referential attribute to determine whether the vessel is navigating or in fishing operation. Since more and more fishing vessels install Automatic Identification System (AIS) either voluntarily or under regulatory requirement, data collected from AIS in real time provide more attributes than SOG which may be utilized to improve the prediction. To be specific, the ships' trajectory patterns and the changes in course become available and should be considered. This paper aims to improve the accuracy in the identification of fishing activities. First, we do feature extraction from the AIS data of coastal waters around Taiwan and build a Recurrent Neural Network (RNN) model. Then, the activity data of fishing vessels are divided into fishing and non-fishing. Finally, based on the testing by feeding various fishing activity data, we can identify the fishing status automatically.
Źródło:
TransNav : International Journal on Marine Navigation and Safety of Sea Transportation; 2020, 14, 3; 527-531
2083-6473
2083-6481
Pojawia się w:
TransNav : International Journal on Marine Navigation and Safety of Sea Transportation
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Forecasting future values of time series using the lstm network on the example of currencies and WIG20 companies
Prognozowanie przyszłych wartości szeregów czasowych z wykorzystaniem sieci lstm na przykładzie kursów walut i spółek WIG20
Autorzy:
Mróz, Bartosz
Nowicki, Filip
Powiązania:
https://bibliotekanauki.pl/articles/2016294.pdf
Data publikacji:
2020
Wydawca:
Politechnika Bydgoska im. Jana i Jędrzeja Śniadeckich. Wydawnictwo PB
Tematy:
recurrent neural network
RNN
gated recurrent unit
GRU
long short-term memory
LSTM
rekurencyjna sieć neuronowa
blok rekurencyjny
pamięć długookresowa
Opis:
The article presents a comparison of the RNN, GRU and LSTM networks in predicting future values of time series on the example of currencies and listed companies. The stages of creating an application which is a implementation of the analyzed issue were also shown – the selection of networks, technologies, selection of optimal network parameters. Additionally, two conducted experiments were discussed. The first was to predict the next values of WIG20 companies, exchange rates and cryptocurrencies. The second was based on investments in cryptocurrencies guided solely by the predictions of artificial intelligence. This was to check whether the investments guided by the predictions of such a program have a chance of effective earnings. The discussion of the results of the experiment includes an analysis of various interesting phenomena that occurred during its duration and a comprehensive presentation of the relatively high efficiency of the proposed solution, along with all kinds of graphs and comparisons with real data. The difficulties that occurred during the experiments, such as coronavirus or socio-economic events, such as riots in the USA, were also analyzed. Finally, elements were proposed that should be improved or included in future versions of the solution – taking into account world events, market anomalies and the use of supervised learning.
W artykule przedstawiono porównanie sieci RNN, GRU i LSTM w przewidywaniu przyszłych wartości szeregów czasowych na przykładzie walut i spółek giełdowych. Przedstawiono również etapy tworzenia aplikacji będącej realizacją analizowanego zagadnienia – dobór sieci, technologii, dobór optymalnych parametrów sieci. Dodatkowo omówiono dwa przeprowadzone eksperymenty. Pierwszym było przewidywanie kolejnych wartości spółek z WIG20, kursów walut i kryptowalut. Drugi opierał się na inwestycjach w kryptowaluty, kierując się wyłącznie przewidywaniami sztucznej inteligencji. Miało to na celu sprawdzenie, czy inwestowanie na podstawie przewidywania takiego programu pozwala na efektywne zarobki. Omówienie wyników eksperymentu obejmuje analizę różnych ciekawych zjawisk, które wystąpiły w czasie jego trwania oraz kompleksowe przedstawienie relatywnie wysokiej skuteczności proponowanego rozwiązania wraz z wszelkiego rodzaju wykresami i porównaniami z rzeczywistymi danymi. Analizowano również trudności, które wystąpiły podczas eksperymentów, takie jak koronawirus, wydarzenia społeczno-gospodarcze czy zamieszki w USA. Na koniec zaproponowano elementy, które należałoby ulepszyć lub uwzględnić w przyszłych wersjach rozwiązania, uwzględniając wydarzenia na świecie, anomalie rynkowe oraz wykorzystanie uczenia się nadzorowanego.
Źródło:
Zeszyty Naukowe. Telekomunikacja i Elektronika / Uniwersytet Technologiczno-Przyrodniczy w Bydgoszczy; 2020, 24; 13-30
1899-0088
Pojawia się w:
Zeszyty Naukowe. Telekomunikacja i Elektronika / Uniwersytet Technologiczno-Przyrodniczy w Bydgoszczy
Dostawca treści:
Biblioteka Nauki
Artykuł

Ta witryna wykorzystuje pliki cookies do przechowywania informacji na Twoim komputerze. Pliki cookies stosujemy w celu świadczenia usług na najwyższym poziomie, w tym w sposób dostosowany do indywidualnych potrzeb. Korzystanie z witryny bez zmiany ustawień dotyczących cookies oznacza, że będą one zamieszczane w Twoim komputerze. W każdym momencie możesz dokonać zmiany ustawień dotyczących cookies