Showing papers on "Hybrid neural network published in 2009"

Extracting rules for classification problems: AIS based approach

Abstract: Although Artificial Neural Network (ANN) usually reaches high classification accuracy, the obtained results in most cases may be incomprehensible. This fact is causing a serious problem in data mining applications. The rules that are derived from ANN are needed to be formed to solve this problem and various methods have been improved to extract these rules. In our previous work, a hybrid neural network was presented for classification (Kahramanli & Allahverdi, 2008). In this study a method that uses Artificial Immune Systems (AIS) algorithm has been presented to extract rules from trained hybrid neural network. The data were obtained from the University of California at Irvine (UCI) machine learning repository. The datasets are Cleveland heart disease and Hepatitis data. The proposed method achieved accuracy values 96.4% and 96.8% for Cleveland heart disease dataset and Hepatitis dataset respectively. It is been observed that these results are one of the best results comparing with results obtained from related previous studies and reported in UCI web sites.

A Hybrid Neural Network and Immune Algorithm Approach for Fit Garment Design

Abstract: The purpose of this study was to develop a system to utilize the successful experiences and help the beginners of garment pattern design (GPD) through optimization methods. Size design of fit garme...

Improved understorey bamboo cover mapping using a novel hybrid neural network and expert system

Abstract: The giant panda is an obligate bamboo grazer. Therefore, the availability and abundance of understorey bamboo determines the quantity and quality of panda habitat. However, there is little or no information about the spatial distribution or abundance of bamboo underneath the forest canopy, due to the limitations of traditional remote sensing classification techniques. In this paper, a new method combines an artificial neural network and a GIS expert system in order to map understorey bamboo in the Qinling Mountains of south-western China. Results from leaf-off ASTER imagery, using a neural network and an expert system, were evaluated for their suitability to quantify understorey bamboo. Three density classes of understorey bamboo were mapped, first using a neural network (overall accuracy 64.7%, Kappa 0.45) and then using an expert system (overall accuracy 62.1%, Kappa 0.43). However, when using the results of the neural network classification as input into the expert system, a significantly improved mapping accuracy was achieved with an overall accuracy of 73.8% and Kappa of 0.60 (average z-value = 3.35, p = 0.001). Our study suggests that combining a neural network with an expert system makes it possible to successfully map the cover of understorey species such as bamboo in complex forested landscapes (e.g. coniferous-dominated and dense canopy forests), and with higher accuracy than when using either a neural network or an expert system.

Study of Genetic Algorithm to Fully-automate the Design and Training of Artificial Neural Network

Abstract: †Summary Optimization of artificial neural network (ANN) parameters design for full-automation ability is an extremely important task, therefore it is challenging and daunting task to find out which is effective and accurate method for ANN prediction and optimization. This paper presents different procedures for the optimization of ANN with aim to: solve the time-consuming of learning process, enhancing generalizing ability, achieving robust and accurate model, and to reduce the computational complexity. A Genetic Algorithm (GA) has been used to optimize operational parameters (input variables), and we plan to optimize neural network architecture (i.e. number of hidden layer and neurons per layer), weight, types, training algorithms, activation functions, learning rate, momentum rate, number of iterations, and dataset partitioning ratio. A hybrid neural network and genetic algorithm model for the determination of optimal operational parameter settings based on the proposed approach was developed. The preliminary result of the model has indicated that the new model can optimize operational parameters precisely and quickly, subsequently, satisfactory performance.

Evolutionary Fuzzy Hybrid Neural Network for Conceptual Cost Estimates in Construction Projects

Abstract: Conceptual cost estimates are important to project feasibility studies, even the final project success. The estimates provide significant information for project evaluations, engineering designs, cost budgeting and cost management. This study proposes an artificial intelligence approach, the evolutionary fuzzy hybrid neural network (EFHNN), to improve precision of conceptual cost estimates. The approach incorporates neural networks (NN) and high order neural networks (HONN) into a hybrid neural network (HNN). The HNN operates with the alternative of linear and non-linear neuron layer connectors. Besides, fuzzy logic (FL) is employed for handling uncertainties, the approach therefore evolve into a fuzzy hybrid neural network (FHNN). For FHNN optimization, the genetic algorithm is used for both FL and HNN, consequently the approach is named as EFHNN. In practical case studies, two estimates including overall and category cost estimates are provided and compared. Results show that the proposed conceptual cost estimates can be deployed as accurate cost estimators during early stages of construction projects. Moreover, considering linear and non-linear neuron layer connectors in EFHNN surpasses models with singular linear deployment of NN.

An enhanced Counter Propagation Neural Network for abnormal retinal image classification

Abstract: Artificial Neural Networks (ANN) is gaining significant importance for pattern recognition applications particularly in the medical field. A hybrid neural network such as Counter Propagation Neural Network (CPN) is highly desirable since it comprises the advantages of supervised and unsupervised training methodologies. Even though it guarantees high accuracy, the network is computationally non-feasible. This drawback is mainly due to the high convergence time period. In this paper, a modified Counter Propagation Neural Network is proposed to tackle this problem which eliminates the iterative training methodology which accounts for the high convergence time. To prove the efficiency, this technique is employed on abnormal retinal image classification system. Real time images from four abnormal classes are used in this work. An extensive feature vector is framed from these images which forms the input for the CPN and the modified CPN. The experimental results of both the networks are analyzed in terms of classification accuracy and convergence time period. The results suggest the superior nature of the proposed technique in terms of convergence time period and classification accuracy.

A Hybrid Neural Network Method for UAV Attack Route Integrated Planning

Abstract: This paper proposes a hybrid neural network method to solve the UAV attack route planning problem considering multiple factors. In this method, the planning procedure is decomposed by two planners: penetration planner and attack planner. The attack planner determines a candidate solution set, which adopts Guassian Radial Basis Function Neural Networks (RBFNN) to give a quick performance evaluation to find the optimal candidate solutions. The penetration planner adopts an alterative Hopfield Neural Network (NN) to refine the candidates in a fast speed. The combined effort of the two neural networks efficiently relaxes the coupling in the planning procedure and is able to generate a near-optimal solution within low computation time. The algorithms are simple and can easily be accelerated by parallelization techniques. Detailed experiments and results are reported and analyzed.

Usage of Hybrid Neural Network Model MLP-ART for Navigation of Mobile Robot

Abstract: We suggest to apply the hybrid neural network based on multi layer perceptron (MLP) and adaptive resonance theory (ART-2) for solving of navigation task of mobile robots. This approach provides semi supervised learning in unknown environment with incremental learning inherent to ART and capability of adaptation to transformation of images inherent to MLP. Proposed approach is evaluated in experiments with program model of robot.

A hybrid neural network strategy for the identification of structural damage using time domain responses

Abstract: A multistage identification scheme for structural damage detection using time domain acceleration responses is proposed. Previous studies of damage assessment using neural networks mostly involved training a backpropagation neural network (BPN) to learn damage patterns with significant computational effort. A hybrid neural network method has been proposed that uses a counterpropagation neural network (CPN) in the first stage for sorting the training data into clusters, giving an approximate guess of the damage extent quickly. After an approximate estimate is obtained, a new set of training patterns of reduced size is generated using the CPN prediction. In the second stage, a BPN trained with the Levenberg–Marquardt algorithm is used to learn the new training data and predict a more accurate result. A superior convergence and a substantial decrease in central processing unit time have been observed for three numerical examples. These examples show the computational superiority of the hybrid method compared...

Making trading decisions for financial-engineered derivatives: a novel ensemble of neural networks using information content

Abstract: Over the last decades, there has been a growing interest in applying artificial intelligence techniques to solve a spectrum of financial problems. A number of studies have shown promising results in using artificial neural networks (ANNs) to guide investment trading. Given the expanding role of ANNs in financial trading, this paper proposes the use of a hybrid neural network, which consists of two independent ANN architectures, and comparatively evaluates its performance against independent ANNs and econometric models in the trading of a financial-engineered (synthetic) derivative composed of options on foreign exchange futures. We examine the financial profitability and the market timing ability of the competing neural network models and statistically compare their attributes with those based on linear and nonlinear statistical projections. A random walk model and the option pricing method are also included as benchmarks for comparison. Our empirical investigation finds that, for each of the currencies analysed, trading strategies guided by the proposed dual network are financially profitable and yield a more stable stream of investment returns than the other models. Statistical results strengthen the notion that diffusion of information contents and cross-validation between the independent components within the dual network are able to reduce bias and extreme decision making over the long run. Moreover, the results are robust with respect to different levels of transaction costs. Copyright © 2009 John Wiley & Sons, Ltd.

Option Price Forecasting Model by Applying Hybrid Neural Network and Genetic Algorithm

Abstract: Implied volatility is the volatility implied by an option price observed in the market.In this paper,A weighted implied volatility measure is regarded as one input of hybrid neural network.We build a hybrid option price forecasting model applying the hybrid neural network and genetic algorithm.The genetic algorithm is applied to the structure optimization of the hybrid neural network and acquisition of the optimal weight of the implied volatility.Case study on Hong Kong derivative market shows that the hybrid forecasting model is better than the conventional Black-Scholes model.

Intelligent modelling of a batch reactor with partially unmeasurable states based upon a structure approaching hybrid neural networks

Abstract: The success of model-based control of chemical processes is dependent on good process models. For processes that are poorly known, the generic modelling capability of neural networks offers an attractive alternative. However, for satisfactory performance, the conventional implementations of neural networks require large sets of offline data in addition to online measurement of key variables, such as concentrations. Meeting each of these requirements is often infeasible in chemical processes. By combining the structural information from a first-principles model and the virtual supervisor-artificial immune algorithm, a novel hybrid neural network, called a structure approaching hybrid neural networks (SAHNN), is proposed. The proposed approach solves the structural problem of neural models and requires a more manageable number of offline data and online key variables. The accurate prediction of online partially unmeasurable concentrations in a batch reactor demonstrated that SAHNN is a promising tool to model complicated batch processes and can be utilized as a vehicle for the control and optimization of other similar chemical reactors.

RepTor: An Intelligent Hybrid Neural Network Based Recipe Generator for Semiconductor Process Modeling and Characterization

Abstract: An intelligent hybrid neural network based recipe generator is presented as a convenient tool for process optimization typically aiming highly-nonlinear plasma process in semiconductor manufacturing. As the wafer size continuously expanding up to 300mm in current high-volume manufacturing (even forecasting 450mm in 2012), fast and convenient process settlement cannot be over emphasized to meet the time-to-market of the newly developed products. In this paper, we suggest a recipe generator based on neural network and particle swamp optimization, nominally RepTor, to help minimizing material cost for process setup and maximizing accuracy of process modeling. RepTor is verified using SiO2 deposition process for modeling and predicting the wafer geometry in conjunction with tool parameter. We have convinced the capability of the suggested recipe generator, and it provides a good starting point for further fine tuning of process optimization.

Dynamics of Incremental Learning by VSF-Network

Abstract: In this paper, we report the dynamics of VSF-Network. VSF-Network is a neural network for the incremental learning and it is a hybrid neural network combining the chaos neural network with a hierarchical network. VSF-Network can find the unknown elements from input with clusters generated by the chaos neuron. We introduce new incremental learning model to explain the dynamics of VSF-Network in this paper. We show the result of analysis of the dynamics of VSF-Network. In the analysis, we focused on the connection weights between layers and neuron cluster generated by the chaotic behavior.

Researches on Intrusion Detection System Based on RBF and Elman Hybrid Neural Network

Abstract: This paper proposes an intrusion detection model based on RBF and Elman Hybrid Neural Network model. This model has the memory function,can effective detection discrete and linked to attack. This model has the memory function,can effective detection discrete and linked to attack. The RBF network is a real-time mode classifier,and the Elman network can remember events. This system USES DARPA data set to test and evaluation. Use ROC curves to display of the test results. The experiment proves this system can effectively improve the detection rate,reduce false alarm rate and missed alarm rate.

Hybrid Neural Network-Based Fault Diagnosis and Fault-Tolerance Design with Application in Electro-Hydraulic Servovalve

Abstract: In this study, to cope with the needs of the predictive maintenance for complex systems, a hybrid dynamic Arti- ficial Neural Network (ANN) based fault and degradation diagnosis and tolerance method is designed. The multi-layer feed forward ANN and recurrent ANN are combined, so as to be able to form a dynamic identification model for the non- linear time-varying system. It has three work modes, and can perform the fault and degradation diagnosis and tolerance by using these modes alternately. The result of its application in an Electro-Hydraulic Servovalve of a Hydroelectric Genera- tion Unit shows that it is effective and feasible, has the advantages of the simple and fast algorithms, working online, and no disturbance signals importing to the system.

Facial Expression Analysis - A Hybrid Neural Network Based Approach

Abstract: It is argued that for the computer to be able to interact with humans, it needs to have human communication skills. One of these skills is the ability to understand the emotional state of human. This paper describes neural network based approaches for emotion classification. We learn a classifier that can recognize 6 basic emotions with an average accuracy of 83% over the Cohn-Kanade database. The novelty of this work is that instead of empirically selecting the parameters of the neural network, i.e. the learning rate, activation function parameter, momentum number, etc., we developed a strategy that can automatically select comparatively better combination of these parameters. We also introduce another way to perform back propagation. Instead of using the error function to adjust the weights of a neural network, we use optimal algorithms such as Powell algorithm to minimize the error function. We also perform several experiments and show that our hybrid neural network approach can be successfully used for emotion recognition.

A hybrid neural network and freight volumes application

Abstract: Since the BP neural network algorithm has some unavoidable disadvantages, such as slowly converging speed and easily running into local minimum, the genetic algorithm and simulated annealing algorithm with the overall search capability have been put forward to optimize authority value and threshold value of BP nerve network. In this paper, a new neural network model which is optimized by genetic algorithm and simulated annealing algorithm has been established and applied into the freight volumes forecast. The result shows that the optimized neural network has significant advantages inspect of fast convergence speed, good generalization ability and not easy to yield minimal local results. In generally, the optimized neural network exhibits good representation and strong prediction ability, and is a helpful tool in the future freight volumes prediction.

Hybrid Neural Network Architecture Based on Self-Organizing Feature Maps

Abstract: An SOM(self-organizing feature maps)-based integrated network,namely ICBP-SOM,is constructed by applying the ICBP network model to the BP-SOM architecture.BP-SOM is a learning algorithm put forward by Ton Weijters,which aims to overcome some of the serious limitations of BP in generalizing knowledge from certain types of learning material.The motivation of presenting the integration is to employ BP-SOM good knowledge interpretation ability and the ICBP good generalization and adaptability to construct an ICBP-SOM,which processes favorable knowledge representation capability and competitive generalization performance.The experimental results on six benchmark data sets validate the feasibility and effectiveness of the integration.

Diaphragm wall's deformation forecasting based on bp-rbf neural networks

Abstract: A artificial neural network is adopted to forecast diaphragm wall's deformations.Five parameters,the soil's cohesion C,the soil's internal friction angle ,the wall's height H,the excavation depth H1 and the survey point's depth h,governing diaphragm wall's deformation are abstracted and taken as inputs of the artificial neural network model.A new hybrid neural network model,BP-RBF Neural Network Model is established by combining the traditional BP and RBF neural network.This new neural network model shows great superiority in higher efficiency and a simpler network structure compared with the traditional pure BP neural network model,at the same time the forecasting accuracy is ensured.

Modeling of Distillation Column Based on Hybrid Neural Network and Prior Knowledge

Abstract: The process of ethanol distillation is a complex chemical process with slowness of dynamic process,complexity of inner mechanism and coupling parameters. For distillation column of ethanol,a hybrid model based on prior knowledge model and neural network model is built. The accuracy of prior knowledge model and range of neural network model. as well as the dependable advanced model for advanced control is improved. The simulation results show that this model has enhanced the performance. The advanced control algorithm of distillation column based on the hybrid model is the future work.

Wavelets Hybrid Neural Network for Harmonic Sources Recognition with the V-I Characteristics

Abstract: This paper presents a wavelets hybrid neural network (WHNN) for harmonic source recognition with voltage-current (V-I) characteristics. With time-domain technique, WHNN is applied in the application of harmonic sources classification. The proposed hybrid network is a two-subnetwork architecture, consisting of self-organizing feature lattice network (SOFL) and wavelet layer connected in cascade. The first Layer Morlet wavelets, as an extractor, are used to extract the features from voltages and currents, and SOFL is employed to classify the various feature patterns including electronic devices, DC/AC motor, and electric arc furnaces (EAFs) in second layer. The computational results have show that With the time-domain technique, the effectiveness of the proposed model is demonstrated by tests. The recognition results of different harmonic sources show the computational efficiency and accurate recognition.

Drug-like and non drug-like pattern classification based on simple topology descriptor using hybrid neural network.

Abstract: An intelligent prediction system has been developed to discriminate drug-like and non drug-like molecules pattern. The system is constructed by using the application of advanced version of standard multilayer perceptron (MLP) neural network called Hybrid Multilayer Perceptron (HMLP) neural network and trained using Modified Recursive Prediction Error (MRPE) training algorithm. In this work, a well understood and easy excess Rule of Five + Veber filter properties are selected as the topological descriptor. The main idea behind the selection of this simple descriptor is to assure that the system could be used widely, beneficial and more advantageous regardless at all user level within a drug discovery organization.