A Modified Conjugate Gradient Formula for Back Propagation Neural Network Algorithm

Al-Bayati, Abbas Y.; Sulaiman, Najmaddin A.; Sadiq, Gulnar W.

doi:10.3844/jcssp.2009.849.856

Cited by 15 publications

(8 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…At this stage, the ANN is considered trained. The backpropagation algorithm based upon the generalized delta rule, proposed firstly by Rumelhart et al (1986) and later by Al Bayati et al (2009), was used to train the ANN in the present study. In the back-propagation algorithm, a set of inputs and outputs is selected from the training set and the network calculates the output based on the inputs.…”

Section: Artificial Neural Networkmentioning

confidence: 99%

Prediction of water flows in Colorado River, Argentina

Pierini

Gómez

Telesca

2012

lajar

View full text Add to dashboard Cite

ABSTRACT. The identification of suitable models for predicting daily water flow is important for planning and management of water storage in reservoirs of Argentina. Long-term prediction of water flow is crucial for regulating reservoirs and hydroelectric plants, for assessing environmental protection and sustainable development, for guaranteeing correct operation of public water supply in cities like Catriel, 25 de Mayo, Colorado River and potentially also Bahía Blanca. In this paper, we analyze in Buta Ranquil flow time series upstream reservoir and hydroelectric plant in order to model and predict daily fluctuations. We compare results obtained by using a three-layer artificial neural network (ANN), and an autoregressive (AR) model, using 18 years of data, of which the last 3 years are used for model validation by means of the root mean square error (RMSE), and measure of certainty (Skill). Our results point out to the better performance to predict daily water flow or refill them of the ANN model performance respect to the AR model. Keywords: prediction, time series, neural networks, autoregressive models, flows, Colorado River, Argentina. Predicción de caudales en río Colorado, ArgentinaRESUMEN. La identificación de modelos adecuados para predecir caudales diarios es importante para la planificación y la gestión de almacenamiento de agua en los embalses de la Argentina. La predicción a largo plazo del caudal es crucial para la regulación de los embalses y centrales hidroeléctricas, evaluar la protección del medio ambiente y el desarrollo sostenible, garantizar el correcto funcionamiento del abastecimiento público de agua en ciudades como Catriel, 25 de Mayo, río Colorado y también, eventualmente, en Bahía Blanca. En este trabajo, se analizan series de tiempo de caudales de agua, arriba del embalse y de la planta hidroeléctrica en Buta Ranquil, para modelar y predecir las fluctuaciones diarias. Se comparan los resultados obtenidos mediante el uso de una red neuronal artificial (ANN) de tres capas y un modelo autoregresivo (AR), con 18 años de datos, cuyos últimos 3 años se utilizan para la validación del modelo por medio de la raíz del error cuadrado medio (RMSE) y medida de certeza (Skill). Para predecir o rellenar el caudal diario, los resultados indican que el mejor desempeño es del ANN con respecto al modelo AR.

show abstract

Section: Artificial Neural Networkmentioning

confidence: 99%

Prediction of water flows in Colorado River, Argentina

Pierini

Gómez

Telesca

2012

lajar

View full text Add to dashboard Cite

show abstract

“…Neural network: A mathematical model which is encouraged by the structure and functional features of biological neural networks is called as artificial neural network (Cybenko, 1989;Hornik, 1991;Wan, 1990;Bayati et al, 2009) which is generally called as neural network. The general architecture of the neural network is given in Fig.…”

Section: Identification Of Ground Water Using Ai Techniquesmentioning

confidence: 99%

The Development of Geo-Information System for Finding Potential Zones of Water using Environmental Parameters and Engineering

Fadi¹

2012

American Journal of Applied Sciences

View full text Add to dashboard Cite

Problem statement: Water is a major natural resource for all the living beings in the world. Peopleâs lives and livelihood depends on water. Most of the human beings in this world use ground water for drinking purpose. The reason behind that is the ground water is pollution less or less polluted when compared with the surface water. The need for clean water increases continually with the world population growth. People in different areas in this world are lack of fresh drinkable water which is important for their survival. Maintaining secure water supplies for drinking, industry and agriculture is not possible without ground water. So it is necessary to explore the potential ground water area to dig a well for the utilization of ground water. Approach: In this study, we have used some parameters to identify the level of water in a particular area. Then, the ground water identification system is designed with the help of Histogram Equalization, Neural Network and PCA. The collected data is given as input to data normalization and the feature is computed for every data using PCA. The presence of ground water in particular location is identified using the trained neural network. Results: Finally, the experimentation is carried out using the synthetic data to show the performance of the ground water identification system. Conclusion: The potential zones of ground water can be identified if the specified parameters of the particular location can be given as input to the neural network

show abstract

“…In [9] an improved BP is proposed where each training pattern has its own activation function of neurons in hidden layer to avoid local minima Wang et al have proposed an individual inference adjusting learning rate technique to enhance the learning performance of the BP neural network [10]. In [11] Conjugate Gradient (CG) algorithm which is usually used for solving nonlinear functions is combined with the modified Back Propagation (BP) algorithm yielding a new fast training multilayer algorithm. The proposed algorithm improved the training efficiency of BP-NN algorithms by adaptively modifying the initial search direction.…”

Section: Introductionmentioning

confidence: 99%

Improved Back Propagation Algorithm to Avoid Local Minima in Multiplicative Neuron Model

Burse¹,

Manoria²,

Kirar³

2011

Information Technology and Mobile Communication

View full text Add to dashboard Cite

The back propagation algorithm calculates the weight changes of artificial neural networks, and a common approach is to use a training algorithm consisting of a learning rate and a momentum factor. The major drawbacks of above learning algorithm are the problems of local minima and slow convergence speeds. The addition of an extra term, called a proportional factor reduces the convergence of the back propagation algorithm. We have applied the three term back propagation to multiplicative neural network learning. The algorithm is tested on XOR and parity problem and compared with the standard back propagation training algorithm.

show abstract

A Modified Conjugate Gradient Formula for Back Propagation Neural Network Algorithm

Cited by 15 publications

References 14 publications

Prediction of water flows in Colorado River, Argentina

Prediction of water flows in Colorado River, Argentina

The Development of Geo-Information System for Finding Potential Zones of Water using Environmental Parameters and Engineering

Improved Back Propagation Algorithm to Avoid Local Minima in Multiplicative Neuron Model

Contact Info

Product

Resources

About