scholarly journals Learning of N-layers neural network

2005 ◽  
Vol 53 (6) ◽  
pp. 75-84 ◽  
Author(s):  
Vladimír Konečný ◽  
Anděla Matiášová ◽  
Ivana Rábová
Keyword(s):  
Artificial Intelligence ◽  
Neural Networks ◽  
Learning Process ◽  
Process Management ◽  
Error Function ◽  
Back Propagation ◽  
Neural Net ◽  
Learning Factor ◽  
The Stability ◽  
Quality Decision

In the last decade we can observe increasing number of applications based on the Artificial Intelligence that are designed to solve problems from different areas of human activity. The reason why there is so much interest in these technologies is that the classical way of solutions does not exist or these technologies are not suitable because of their robustness. They are often used in applications like Business Intelligence that enable to obtain useful information for high-quality decision-making and to increase competitive advantage.One of the most widespread tools for the Artificial Intelligence are the artificial neural networks. Their high advantage is relative simplicity and the possibility of self-learning based on set of pattern situations.For the learning phase is the most commonly used algorithm back-propagation error (BPE). The base of BPE is the method minima of error function representing the sum of squared errors on outputs of neural net, for all patterns of the learning set. However, while performing BPE and in the first usage, we can find out that it is necessary to complete the handling of the learning factor by suitable method. The stability of the learning process and the rate of convergence depend on the selected method. In the article there are derived two functions: one function for the learning process management by the relative great error function value and the second function when the value of error function approximates to global minimum.The aim of the article is to introduce the BPE algorithm in compact matrix form for multilayer neural networks, the derivation of the learning factor handling method and the presentation of the results.

A Reverse Approach in Optimizing Pass Parameters

2010 ◽  
Vol 113-116 ◽  
pp. 1707-1711
Author(s):  
Jian Hua Hu ◽  
Yuan Hua Shuang
Keyword(s):  
Neural Networks ◽  
Finite Element Method ◽  
Finite Element ◽  
Learning Process ◽  
Back Propagation ◽  
Fem Simulation ◽  
Steel Pipe ◽  
Good Convergence ◽  
Back Propagation Neural Networks ◽  
Element Method

A method combines a back propagation neural networks (BPNN) with the data obtained using finite element method (FEM) is introduced in this paper as an approach to solve reverse problems. This paper presents the feasibility of this approach. FEM results are used to train the BPNN. Inputs of the network are associated with dimension deviation values of the steel pipe, and outputs correspond to its pass parameters. Training of the network ensures low error and good convergence of the learning process. At last, a group of optimal pass parameters are obtained, and reliability and accuracy of the parameters are verified by FEM simulation.

Applying Neural Networks for Modeling of Financial Assets

2018 ◽  
pp. 172-204
Author(s):  
Dmitry Averchenko ◽  
Artem Aldyrev
Keyword(s):  
Neural Networks ◽  
Artificial Neural Networks ◽  
Standard Deviation ◽  
Learning Process ◽  
Moving Average ◽  
Back Propagation ◽  
Financial Assets ◽  
Movement Index ◽  
Analytical System ◽  
Feedback Neural Networks

The purpose of this chapter is to develop an analytical system for forecasting prices of financial assets with the use of artificial neural networks technology. Proposed by the authors, the analytical system consists of several neural networks, each of which makes the forecast of financial assets prices. The system includes recurrence (with feedback) neural networks with sigmoidal activation formula. This allows the networks to “remember” a sequence of reactions to the same stimulus. The learning process of neural networks is performed using an algorithm of back propagation of error. The key parameters of forecast for this analytical system are the indicators presented by the terminal MetaTrader 4-broker Forex Club: Average Directional и Movement Index; Bollinger Bands; Envelopes; Ichimoku Kinko Hyo; Moving Average; Parabolic SAR; Standard Deviation; Average True Range; and others.

scholarly journals Comparison of Artificial Intelligence Methods for Fault Classification of the 115-kV Hybrid Transmission System

2020 ◽  
Vol 10 (11) ◽  
pp. 3967 ◽  
Author(s):  
Jittiphong Klomjit ◽  
Atthapol Ngaopitakkul
Keyword(s):  
Artificial Intelligence ◽  
Neural Networks ◽  
Transmission Line ◽  
Fault Location ◽  
Back Propagation ◽  
Fault Classification ◽  
Support Vector ◽  
Discrete Wavelet ◽  
Mother Wavelet ◽  
Hybrid Transmission

This research proposes a comparison study on different artificial intelligence (AI) methods for classifying faults in hybrid transmission line systems. The 115-kV hybrid transmission line in the Provincial Electricity Authority (PEA-Thailand) system, which is a single circuit single conductor transmission line, is studied. Fault signals in the transmission line were generated by the EMTP/ATPDraw software. Various factors such as fault location, type, and angle were considered. Then, fault signals were analyzed by coefficient details on the first scale of the discrete wavelet transform. Daubechies mother wavelet from MATLAB software was used to decompose the fault signal. The coefficient value of the mother wavelet behaved depending on the position, inception of fault angle, and fault type. AI methods including probabilistic neural networks (PNNs), back-propagation neural networks (BPNNs), and support vector machine (SVM) were used to identify faults. AI input used the maximum first peak coefficients of phase ABC and zero sequence. The results obtained from the study were found to be satisfactory with all AI methodologies having an average accuracy of more than 98% in the case study. However, the SVM technique can provide more accurate results than the PNN and BPNN techniques with less computation burden. Thus, it is suitable for being applied to actual protection systems.

Improved Artificial BP Neural Networks and Genetic Algorithm Used to Predict the Purity of the Artificial Synthetic Hydrotalcite

2007 ◽  
Vol 280-283 ◽  
pp. 495-498
Author(s):  
Qiang Luo ◽  
Qing Li Ren
Keyword(s):  
Genetic Algorithm ◽  
Neural Networks ◽  
Prediction Model ◽  
Process Parameters ◽  
Learning Algorithm ◽  
Back Propagation ◽  
Raw Material ◽  
Algorithm Analysis ◽  
Neural Net ◽  
Bp Neural Networks

A prediction model for purity of the artificial synthetic hydrotalcite under varied process parameters based on improved artificial back-propagation (BP) neural networks is developed. And the non-linear relationship between the hydrotalcite purity and the raw material adding amount of NaOH, MgCl2 and AlCl3 was established based on BP learning algorithm analysis and convergence improvement. The hydrotalcite purity can be predicted by means of the trained neural net. Thus, by virtue of the prediction model, the future hydrotalcite purity can be evaluated under random complicated raw material amounts. Moreover, the best processing technology is optimized using the genetic algorithm.

scholarly journals Modelling an Adaptive Learning System Using Artificial Intelligence

Webology ◽  
2021 ◽  
Vol 19 (1) ◽  
pp. 01-18
Author(s):  
Hayder Rahm Dakheel AL-Fayyadh ◽  
Salam Abdulabbas Ganim Ali ◽  
Dr. Basim Abood
Keyword(s):  
Artificial Intelligence ◽  
Neural Networks ◽  
Adaptive Learning ◽  
Back Propagation ◽  
Learning System ◽  
Spiking Neural Networks ◽  
Domain Specific Language ◽  
Domain Specific ◽  
Adaptive Learning System ◽  
Artificial Neural

The goal of this paper is to use artificial intelligence to build and evaluate an adaptive learning system where we adopt the basic approaches of spiking neural networks as well as artificial neural networks. Spiking neural networks receive increasing attention due to their advantages over traditional artificial neural networks. They have proven to be energy efficient, biological plausible, and up to 105 times faster if they are simulated on analogue traditional learning systems. Artificial neural network libraries use computational graphs as a pervasive representation, however, spiking models remain heterogeneous and difficult to train. Using the artificial intelligence deductive method, the paper posits two hypotheses that examines whether 1) there exists a common representation for both neural networks paradigms for tutorial mentoring, and whether 2) spiking and non-spiking models can learn a simple recognition task for learning activities for adaptive learning. The first hypothesis is confirmed by specifying and implementing a domain-specific language that generates semantically similar spiking and non-spiking neural networks for tutorial mentoring. Through three classification experiments, the second hypothesis is shown to hold for non-spiking models, but cannot be proven for the spiking models. The paper contributes three findings: 1) a domain-specific language for modelling neural network topologies in adaptive tutorial mentoring for students, 2) a preliminary model for generalizable learning through back-propagation in spiking neural networks for learning activities for students also represented in results section, and 3) a method for transferring optimised non-spiking parameters to spiking neural networks has also been developed for adaptive learning system. The latter contribution is promising because the vast machine learning literature can spill-over to the emerging field of spiking neural networks and adaptive learning computing. Future work includes improving the back-propagation model, exploring time-dependent models for learning, and adding support for adaptive learning systems.

scholarly journals FASTER TRAINING USING FUSION OF ACTIVATION FUNCTIONS FOR FEED FORWARD NEURAL NETWORKS

2009 ◽  
Vol 19 (06) ◽  
pp. 437-448 ◽  
Author(s):  
MD. ASADUZZAMAN ◽  
MD. SHAHJAHAN ◽  
KAZUYUKI MURASE
Keyword(s):  
Neural Networks ◽  
Time Series ◽  
Error Function ◽  
Time Series Prediction ◽  
Back Propagation ◽  
Activation Functions ◽  
Classification Problems ◽  
Feed Forward ◽  
Feed Forward Neural Networks ◽  
Prediction Problems

Multilayer feed-forward neural networks are widely used based on minimization of an error function. Back propagation (BP) is a famous training method used in the multilayer networks but it often suffers from the drawback of slow convergence. To make the learning faster, we propose 'Fusion of Activation Functions' (FAF) in which different conventional activation functions (AFs) are combined to compute final activation. This has not been studied extensively yet. One of the sub goals of the paper is to check the role of linear AFs in combination. We investigate whether FAF can enable the learning to be faster. Validity of the proposed method is examined by performing simulations on challenging nine real benchmark classification and time series prediction problems. The FAF has been applied to 2-bit, 3-bit and 4-bit parity, the breast cancer, Diabetes, Heart disease, Iris, wine, Glass and Soybean classification problems. The algorithm is also tested with Mackey-Glass chaotic time series prediction problem. The algorithm is shown to work better than other AFs used independently in BP such as sigmoid (SIG), arctangent (ATAN), logarithmic (LOG).

Soft Sensor Design for Biodiesel Concentration in a Transesterification Reactor

2008 ◽  
Author(s):  
Jaime Garci´a ◽  
Jose´ Posada ◽  
Pedro Villalba ◽  
Marco Sanjuan
Keyword(s):  
Neural Networks ◽  
Process Model ◽  
Continuous Production ◽  
Continuous Process ◽  
Back Propagation ◽  
Alcohol Concentration ◽  
Soft Sensor ◽  
Optimum Number ◽  
Neural Net ◽  
Process Data

Biofuels production is facing new challenges every day, related to better process control and quality monitoring. It is very important for the sustainability of these processes to implement strategies and alternatives in order to achieve a continuous production process and to control significant variables involved in the reaction. One of the most difficult variables to measure is the actual Biodiesel concentration inside the reactor. Neural networks have become a useful strategy to give solutions to complex problems; its application is growing faster at industries due to the inherent nonlinear behavior of the processes, modeled easily by this computational tool. The capacity of mapping a complex behavior trough input and output process data, without a complicated and hardly to obtain mathematical model, makes neural networks an attractive strategy to be implemented in most industries, in a soft sensor or a process model scheme. This investigation addresses the need to predict the concentrations of esters (biodiesel) when different triglycerides are reacting with alcohol. Concentration was estimated using an approach that uses a soft sensor that captures the dynamics of these variables through off line laboratory experiments. The soft sensor is actually a Random Activation Weight Neural Net (RAWN), which is a back propagation neural network with a fast training algorithm that does not need any iteration. Also, to reduce the complexity of the soft sensor an optimization procedure was carried out to determine the optimum number of neurons in the hidden layer. In this research Biodiesel was produced by transesterification of palm oil with ethanol and KOH as catalyst. During transesterification reaction the estimation of concentrations is determined by laboratory analysis at off line stages, these variables are very important to control the continuous process of a biodiesel plant.

scholarly journals Analysis of the ReSuMe Learning Process For Spiking Neural Networks

2008 ◽  
Vol 18 (2) ◽  
pp. 117-127 ◽  
Author(s):  
Filip Ponulak
Keyword(s):  
Neural Networks ◽  
Learning Process ◽  
Learning Algorithm ◽  
Real Life ◽  
Optimal Solution ◽  
Spiking Neural Networks ◽  
Fast Learning ◽  
Life Tasks ◽  
The Neural Networks ◽  
The Stability

Analysis of the ReSuMe Learning Process For Spiking Neural NetworksIn this paper we perform an analysis of the learning process with the ReSuMe method and spiking neural networks (Ponulak, 2005; Ponulak, 2006b). We investigate how the particular parameters of the learning algorithm affect the process of learning. We consider the issue of speeding up the adaptation process, while maintaining the stability of the optimal solution. This is an important issue in many real-life tasks where the neural networks are applied and where the fast learning convergence is highly desirable.

KINK FINDING IN THE ALEPH TPC USING NEURAL NETWORKS

1993 ◽  
Vol 08 (29) ◽  
pp. 2715-2727 ◽  
Author(s):  
GEORG STIMPFL-ABELE
Keyword(s):  
Neural Networks ◽  
Back Propagation ◽  
Simulated Data ◽  
Radial Position ◽  
Neural Net ◽  
Decay Vertex ◽  
Feed Forward ◽  
Error Back Propagation ◽  
Conventional Methods ◽  
Tracking Device

The task of finding the decays of charged tracks inside a tracking device is divided into two parts. First a neural net is used to recognize kinks in well-reconstructed tracks. If a kink has been found, a second net determines the radial position of the decay vertex. Both algorithms use feed-forward nets with error back-propagation. Very good performance is obtained in comparison with conventional methods using simulated data from the ALEPH TPC. The behavior of the nets is analyzed by studying the correlations between the inputs and the output.

scholarly journals Implementasi Jaringan Syaraf Tiruan (JST) untuk Mengenali Pola Tanda Tangan dengan Metode Backpropagation

2020 ◽  
Vol 3 (2) ◽  
pp. 51-57
Author(s):  
Aldo Vyan Martha ◽  
Mukhtar Hanafi ◽  
Auliya Burhanuddin
Keyword(s):  
Artificial Intelligence ◽  
Neural Networks ◽  
Artificial Neural Networks ◽  
Computer Technology ◽  
Learning Process ◽  
Learning Algorithms ◽  
Complex Data ◽  
Sample Data ◽  
Backpropagation Method ◽  
Signature Pattern

Artificial Neural Networks (ANN) is a computer technology in the field of artificial intelligence that is able to understand complex data patterns. One of ANN's technological capabilities is being able to predict solutions based on training patterns provided during the system learning process. This study aims to apply the signature pattern by applying ANN using the Backpropagation method. Backpropagation method is one of the learning algorithms related to the preparation of weights based on the value of errors in learning. The image will be processed using the Backpropagation method which will be obtained by the introduction. The results introduce 50 signature data samples and 50 signature sample data. The test is carried out using 50 samples, where each sample will be requested once. From the results of the research that has been done it can be concluded that the results obtained from the parameters with a learning rate of 0.5, epoch 100, objectives 1e-5 and momentum 0.9 with the results of 68% system testing.

Sign in / Sign up

Export Citation Format

Share Document