Three different learning rnethods for RBF networks and their combinations are preserited. Standard gradient learning, three-step algorithm with unsupervised part, and evolutionary algorithm are introduced. Their performance is compared on two benchmark problerns: Two spirals and Iris plants. The results show that the three-step learning is usually the fastest, while the gradient learning achieves better precision. The cornbination of these two approaches gives the best results.
A new optimization model for determining the minimum cost for the design of water distribution systems is described in the paper based on a combination of linear programming methodology (LP) and a genetic algorithms approach (GA). The optimal design of looped hydraulic pipe networks belongs to the class of large combinatorial optimization problems that are difficult to handle using conventional operational research techniques. A many different heuristic algorithms have been developed in the last two decades and applied to the design of water distribution systems. Although many research efforts have been made for the sake of achieving the optimal design of the looped water distribution networks, there is still some uncertainty about finding a generally reliable method. The authors of the paper are proposing a method in which the main emphasis is placed on its higher reliability of finding an optimal solution in terms of closeness to a global minimum. In this study the efficiency of GA optimization was improved through a hybrid method, which combines the GA method (heuristic component) with a linear programming methodology (deterministic component). The proposed method was tested on three benchmark least-cost design problems and compared with other methods and the results suggest that the GALP consistently provides better solutions. It was investigated that the method gives results more reliable in terms of closeness to a global minimum. The proposed method guarantees very close convergence to a global optimum. The method is intended for use in the design and rehabilitation of drinking water systems and pressurized irrigation systems as well. and V článku je predstavený model cenovej optimalizácie pri navrhovaní priemerov potrubí okruhových tlakových rozvodov určených na distribúciu vody, ktorý je založený na kombinácii metódy lineárneho programovania (LP) a genetických algoritmov (GA). Cenovo optimálny návrh okruhových sietí patrí do triedy relatívne veľkých kombinatorických problémov, ktoré sú ťažko riešiteľné klasickými technikami matematického programovania. Z tohto dôvodu bolo v posledných dvoch dekádach vyvinutých aj viacero heuristických algoritmov. Napriek tomu pretrváva určitá neistota, pokiaľ ide o spoľahlivosť existujúcich metód, ktorá sa odzrkadľuje ich malým využitím v praxi. Autori predkladajú metódu, v ktorej sa pri hľadaní optimálneho návrhu okruhových sústav v zmysle priblíženia sa ku globálnemu minimu hlavný dôraz kladie na vyššiu spoľahlivosť. Ide o hybridnú metódu, ktorá kombinuje genetické algoritmy (heuristická zložka) s lineárnym programovaním (deterministická zložka). Autori verifikujú metódu na benchmarkovej distribučnej sústave, používanej na testovanie v odbornej literatúre a na dvoch modeloch derivovaných z tejto sústavy so známym riešením. Výsledky potvrdzujú, že v porovnaní s existujúcimi modelmi navrhovaná metóda zaručí vyššiu úroveň kvality dosahovaných výsledkov, tak v zmysle priblíženia sa ku globálnemu minimu, ako aj z hľadiska jednoduchšieho nastavovania parametrov GA a rýchlejšej konvergencie. Metódu možno použiť pri navrhovaní nových systémov aj na rekonštrukciu existujúcich systémov na rozvod pitnej vody a pri navrhovaní tlakových závlahových systémov.
Neural network based on back-propagation (BP) algorithm is a widely used prediction model. However, the nodes number of the first hidden layer, the learning rate and momentum factor are usually determined manually, which affects the forecast accuracy of network. Therefore, in this paper, to improve the forecast accuracy, firstly, the nodes number of the first hidden layer is selected adaptively based on minimizing mean square error (MSE). Secondly, improved genetic algorithm (GA) is proposed to train the learning rate and momentum factor dynamically, which includes multi-point crossover and single point mutation. Thirdly, we construct a new neural network model based on the adaptively selected nodes number of the first hidden layer, the dynamically selected learning rate and momentum factor, which is called HN-GA-BP neural network model. Finally, the proposed neural network model is used to forecast the carbon dioxide contents in China for fifty years. Experimental results demonstrate the effectiveness of the proposed HN-GA-BP neural network model.
This paper presents a novel framework to use artificial neural network (ANN) for accurate forecasting of river flows at higher lead times. The proposed model, termed as sequential ANN (SANN), is based on the heuristic that a mechanism that provides an accurate representation of physical condition of the basin at the time of forecast, in terms of input information to ANNs at higher lead time, helps improve the forecast accuracy. In SANN, a series of ANNs are connected sequentially to extend the lead time of forecast, each of them taking a forecast value from an immediate preceding network as input. The output of each network is modified by adding an expected value of error so that the residual variance of the forecast series is minimized. The applicability of SANN in hydrological forecasting is illustrated through three case examples: a hypothetical time series, daily river flow forecasting of Kentucky River, USA and hourly river flow forecasting of Kolar River, India. The results demonstrate that SANN is capable of providing accurate forecasts up to 8 steps ahead. A very close fit (>94% efficiency) was obtained between computed and observed flows up to 1 hour in advance for all the cases, and the deterioration in fit was not significant as the forecast lead time increased (92% at 8 steps ahead). The results show that SANN performs much better than traditional ANN models in extending the forecast lead time, suggesting that it can be effectively employed in developing flood management measures.
This work is motivated by the interest in feature selection that greatly affects the detection accuracy of a classifier. The goals of this paper are (i) identifying optimal feature subset using a novel wrapper based feature selection algorithm called Shapley Value Embedded Genetic Algorithm (SVEGA), (ii) showing the improvement in the detection accuracy of the Artificial Neural Network (ANN) classifier with the optimal features selected, (iii) evaluating the performance of proposed SVEGA-ANN model on the medical datasets. The medical diagnosis system has been built using a wrapper based feature selection algorithm that attempts to maximize the specificity and sensitivity (in turn the accuracy) as well as by employing an ANN for classification. Two memetic operators namely include and remove features (or genes) are introduced to realize the genetic algorithm (GA) solution. The use of GA for feature selection facilitates quick improvement in the solution through a fine tune search. An extensive experimental evaluation of the proposed SVEGA-ANN method on 26 benchmark datasets from UCI Machine Learning repository and Kent ridge repository, with three conventional classifiers, outperforms state-of-the-art systems in terms of classification accuracy, number of selected features and running time.
Intrusion detection systems (IDSs) are designed to distinguish normal and intrusive activities. A critical part of the IDS design depends on the selection of informative features and the appropriate machine learning technique. In this paper, we investigated the problem of IDS from these two perspectives and constructed a misuse based neurotree classiffier capable of detecting anomalies in networks. The major implications of this paper are a) Employing weighted sum genetic feature extraction process which provides better discrimination ability for detecting anomalies in network trafic; b) Realizing the system as a rule-based model using an ensemble efficient machine learning technique, neurotree which possesses better comprehensibility and generalization ability; c) Utilizing an activation function which is targeted at minimizing the error rates in the learning algorithm. An extensive experimental evaluation on a database containing normal and anomaly trafic patterns shows that the proposed scheme with the selected features and the chosen classiffier is a state-of-the-art IDS that outperforms previous IDS methods.
This paper presents a hybrid model based on the displacement back analysis to estimate the earth stress magnitude and direction from the obtained borehole displacements. An artificial neural network (ANN) is used to map the non-linear relationship between the maximum horizontal earth stress, σH, the minimum horizontal earth stress, σh, the direction of the largest horizontal earth stress, θ and the borehole displacements. The genetic algorithm (GA) is used to search the set of unknown earth stresses and direction according to the objective function. Results of the numerical experiments show that the displacement back analysis method can effectively identify the earth stress based on the wellbore motions during drilling. and Obsahuje seznam literatury
The problem of network is formulated as linear programming and genetic algorithm in spreadsheet model. GA’s are based in concept on natural genetic and evolutionary mechanisms working on populations of solutions in contrast to other search techniques that work on a single solution. An example application is presented. An empirical analysis of the effects of the algorithm’s parameters is also Presented in the context of this novel application.
In this paper, we suggest Evolution Algorithms (EA) for development of neural network topologies to find the optimal solution of some problems. Topologies are modified in feed-forward neural networks and in special cases of recurrent neural networks.
We applied two approaches to the tuning of neural networks. One is classical, using evolution principles only. In the other approach, the adaptation phase (training phase) of the neural network is raade in two steps. In the hrst step we use the genetic algorithm to find better than random starting weights (nearly optimal values), in the second step we use the backpropagation algorithm to finish the adaptation phase. This means that the starting weights for the backpropagation algorithm are not random values, but approximately optimum values. In this context, the fitness of a chromosome (neural network) is a function of its estimated test error (its estimated generalization ability).
Some results obtained by these methods are demonstrated in a prediction of Geo-Magnetic Storms (GMS) and Handwriting Recognition (HWR).
Trust-aware recommender system (TARS) recommends ratings based on user trust. It greatly improves the conventional collaborative filtering by providing reliable recommendations when dealing with the data sparseness problem. One basic research issue of TARS is to improve the recommending efficiency, in which the key point is to find sufficient number of recommenders efficiently for active users. Existing works searched recommenders via a skeleton, which consists of a number of hub nodes. The hub nodes are those who have superior degrees based on the scale-freeness of the trust network. However, existing works did not consider the skeleton maintenance cost and the coverage overlap between nodes of the skeleton. They also failed to suggest the proper size of the skeleton. This paper proposes an optimized TARS model to solve the problems of existing works. By using the genetic algorithm, our model chooses the most suitable nodes for the skeleton of recommender searching. It can achieve the maximum prediction coverage with the minimum skeleton maintenance cost. Simulation results show that compared with existing works, our model can reduce more than 90{\%} of the skeleton maintenance cost with reasonable prediction coverage.