Showing papers on "Hybrid neural network published in 2012"

Multi-frame prediction for hybrid neural network/hidden Markov models

Abstract: A method and system for multi-frame prediction in a hybrid neural network/hidden Markov model automatic speech recognition (ASR) system is disclosed. An audio input signal may be transformed into a time sequence of feature vectors, each corresponding to respective temporal frame of a sequence of periodic temporal frames of the audio input signal. The time sequence of feature vectors may be concurrently input to a neural network, which may process them concurrently. In particular, the neural network may concurrently determine for the time sequence of feature vectors a set of emission probabilities for a plurality of hidden Markov models of the ASR system, where the set of emission probabilities are associated with the temporal frames. The set of emission probabilities may then be concurrently applied to the hidden Markov models for determining speech content of the audio input signal.

A Hybrid Neural Network-Latent Topic Model

Abstract: This paper introduces a hybrid model that combines a neural network with a latent topic model. The neural network provides a lowdimensional embedding for the input data, whose subsequent distribution is captured by the topic model. The neural network thus acts as a trainable feature extractor while the topic model captures the group structure of the data. Following an initial pretraining phase to separately initialize each part of the model, a unified training scheme is introduced that allows for discriminative training of the entire model. The approach is evaluated on visual data in scene classification task, where the hybrid model is shown to outperform models based solely on neural networks or topic models, as well as other baseline methods.

A Hybrid Neural Network and ARIMA Model for Energy Consumption Forcasting

Abstract: Energy consumption time series consist of complex linear and non-linear patterns and are difficult to forecast. Neither autoregressive integrated moving average (ARIMA) nor artificial neural networks (ANNs) can be adequate in modeling and predicting energy consumption. The ARIMA model cannot deal with nonlinear relationships while the neural network model alone is not able to handle both linear and nonlinear patterns equally well. In the present study, a hybrid methodology that combines both ARIMA and ANN models is proposed to take advantage of the unique strength of ARIMA and ANN models in linear and nonlinear modeling. The empirical results with energy consumption data of Hebei province in China indicate that the hybrid model can be an effective way to improve the energy consumption forecasting accuracy obtained by either of the models used separately

A hybrid algorithm to optimize cutting parameter for machining GFRP composite using alumina cutting tools

Abstract: In this paper, two different evolutionary algorithm-based neural network models were developed to optimise the unit production cost. The hybrid neural network models are, namely, genetic algorithm-based neural network (GA-NN) model and particle swarm optimization- based neural network (PSO-NN) model. These hybrid neural network models were used to find the optimal cutting conditions of Ti(C,N) mixed alumina-based ceramic cutting tool (CC650) and SiC whisker-reinforced alumina- based ceramic cutting tool (CC670) on machining glass fibre-reinforced plastic (GFRP) composite. The objective considered was the minimization of unit production cost subjected to various machine constraints. An orthogonal design and analysis of variance was employed to determine the effective cutting parameters on the tool life. Neural network helps obtain a fairly accurate prediction, even when enough and adequate information is not available. The GA-NN and PSO-NN models were compared for their performance. Optimal cutting conditions obtained with the PSO-NN model are the best possible compromise com- pared with the GA-NN model during machining GFRP composite using alumina cutting tool. This model also proved that neural networks are capable of reducing uncertainties related to the optimization and estimation of unit production cost.

A review on short term load forecasting using hybrid neural network techniques

Abstract: Load forecasting is very essential for the efficient and reliable operation of a power system. Often uncertainties significantly decrease the prediction accuracy of load forecasting; this in turn affects the operation cost dramatically as well as the optimal day-to-day operation of the power system. In this article, an overview of recently published work on hybrid neural network techniques to forecast the electrical load demand. A hybrid neural network forecasting model is proposed, which is a combination of simulated annealing (SA) and particle swarm optimization (PSO) called SAPSO. In proposed techniqiue, particle swarm optimization (PSO) algorithm has the ability of global optimization and the simulated annealing (SA) algorithm has a strong searching capability. Therefore, the learning algorithm of a typical three layer feed forward neural network back propagation (BP) is replaced by SAPSO algorithm. Furthermore, preprocessing of input data, convergence, local minima and working of neural network with SAPSO algorithm also discussed.

Boosted Hybrid Recurrent Neural Classifier for Text Document Classification on the Reuters News Text Corpus

Abstract: The objective is multi-classed news text classification using hybrid neural techniques on the modapte version of the Reuters news text corpus. In particular, a neuroscience based hybrid neural classifier fully integrated with a novel boosting algorithm is examined for its potential in text document classification in a non-stationary environment. The novel boosting algorithm termed NeuroBoost is an Adaboost-like algorithm that computes and integrates boosted weights into neural network weights, using back-propagation approach. The main contribution of this paper is the provision of an obvious scientific basis for integrating boosted weights into hybrid neural network weights. Results attained in this experiment show impressive performance by the hybrid neural classifier even with minimal number of neurons in constituting structures. A minimal but appreciable increase is observed in performance if an appreciable number of neurons are added.

Training a Feed-Forward Neural Network with Artificial Bee Colony based Backpropagation Method

Abstract: Back-propagation algorithm is one of the most widely used and popular techniques to optimize the feed forward neural network training. Nature inspired meta-heuristic algorithms also provide derivative-free solution to optimize complex problem. Artificial bee colony algorithm is a nature inspired meta-heuristic algorithm, mimicking the foraging or food source searching behaviour of bees in a bee colony and this algorithm is implemented in several applications for an improved optimized outcome. The proposed method in this paper includes an improved artificial bee colony algorithm based back-propagation neural network training method for fast and improved convergence rate of the hybrid neural network learning method. The result is analysed with the genetic algorithm based back-propagation method, and it is another hybridized procedure of its kind. Analysis is performed over standard data sets, reflecting the light of efficiency of proposed method in terms of convergence speed and rate.

Design of an aperture-coupled microstrip antenna using a hybrid neural network

Abstract: In this study, an artificial neural network (ANN) model using hybrid neural network is proposed for the design of aperture-coupled microstrip antennas (ACMSAs). The new hybrid model is developed by combining radial basis function (RBF) and back-propagation algorithm (BPA). The performances evaluation of the hybrid model reveals superiority over the conventional BPA and RBF models in terms of error and time. The results obtained by the proposed model are compared with the simulation results obtained from the IE3D software package and also with the experimental results obtained from the fabricated ACMSA. The results show good agreement.

Ordinal classification using hybrid artificial neural networks with projection and kernel basis functions

Abstract: Many real life problems require the classification of items into naturally ordered classes. These problems are traditionally handled by conventional methods intended for the classification of nominal classes, where the order relation is ignored. This paper proposes a hybrid neural network model applied to ordinal classification using a possible combination of projection functions (product unit, PU) and kernel functions (radial basis function, RBF) in the hidden layer of a feed-forward neural network. A combination of an evolutionary and a gradient-descent algorithms is adapted to this model and applied to obtain an optimal architecture, weights and node typology of the model. This combined basis function model is compared to the corresponding pure models: PU neural network, and the RBF neural network. Combined functions using projection and kernel functions are found to be better than pure basis functions for the task of ordinal classification in several datasets.

Instantaneous pavement condition evaluation using non-destructive neuro-evolutionary approach

Abstract: In this paper, a hybrid neural network (NN)-genetic algorithm (GA) based non-destructive pavement auscultation method for instantaneous airfield infrastructure condition assessment is discussed. NNs are employed for finite element aided forward prediction of pavement surface deflections resulting from non-destructive test impulse loading and the GAs are used for global optimisation of the pavement structural parameters by matching the NN predicted deflections with the measured pavement response. This hybrid approach takes advantage of the non-linear estimation capability provided by neural networks trained using finite element (FE) solutions in modelling the stress-dependent behaviour of unbound pavement geo-materials while improving the robustness to measurement uncertainty through the application of genetic algorithms. The performance of the developed hybrid pavement auscultation technique is evaluated through extensive field studies conducted at a state-of-the-art full-scale airfield pavement test faci...

Control Performance Standard based Load Frequency Control of a two area Reheat Interconnected Power System considering Governor Dead Band nonlinearity using Fuzzy Neural Network

Abstract: The frequency control of reheat interconnected two area power systems are mainly characterized by non-linearity and uncertainty. A hybrid neural network and fuzzy control is proposed for load frequency control in the power systems considering governor dead band (GDB) non-linearity. Fuzzy with neural network is employed to forecast the control input requirement and system's future output, based on the current Area Control Error (ACE) and the predicted change-of-ACE. The Control Performance Standard (CPS) criterion is adopted to the fuzzy controller design, thus improves the dynamic quality of system. The system was simulated and the output responses of frequency deviations in area 1 and area 2 and tie-line power deviations for 1% step-load disturbance in area 1 were obtained. The comparison of frequency deviations and tie-line power deviations for the two area interconnected thermal power system considering GDB nonlinearity with Redox Flow Batteries (RFB) reveals that the system with hybrid fuzzy neural controller enhances a better stability than that of system with integral controller.

Neural Network based Road Sign Recognition

Abstract: A recent surge of interest is to recognize Road Signs. Signs are visual languages that represent some special circumstantial information of environment. They provide important information for guiding, warning people to make their movements safer, easier and more convenient. This thesis presents a hybrid neural network solution for Road sign recognition which combines local image sampling and artificial neural network. The method is based on BAM for dimensional reduction and multi-layer perception with backpropagation algorithm has been used for training the network. It has been found from practical observations that the number of iterations required to train the network is enormous. The capability of recognition of a neural network increases with increasing the training accuracy. For this process each sign is converted to a designated M×N feature matrix. These feature matrices of signs are then fed into the neural network as input patterns. The neural network is trained with the set of input patterns of the digits to acquire separate knowledge corresponding to each Road sign. In order to justify the effectiveness of the system, different test patterns of the signs are used to verify the system. Experimental results demonstrate that the system is capable of recognizing Road signs with 98% accuracy.

Hybrid neural network under-actuated sliding-mode control for trajectory tracking of quad-rotor unmanned aerial vehicle

Abstract: Due to the under-actuated characteristic of quadrotor unmanned aerial vehicle (QUAV), the sliding surface using measurable output (i.e., 3D position and attitude), whose number is larger than that of control input (i.e., total thrust force, roll, pitch and yaw torques), is designed. Hence, the number of control input and sliding surface is the same, and the indirectly controlled mode (e.g., x- and y-axes) is controlled. Under uncertain environment, the sliding-mode under-actuated control (SMUC) with suitable conditions is first derived so that asymptotical and bounded tracking results are achieved. To improve system performance, an on-line recurrent neural network modeling for dynamical uncertainty of QUAV is employed to design a recurrent-neural-network-based sliding-mode under-actuated control (RNNSMUC). Then the proposed hybrid neural-network-based sliding-mode under-actuated control (HNNSMUC) combining SMUC and RNNSMUC with a transition maintains both advantages of SMUC and RNNSMUC and simultaneously avoids the disadvantages coming from SMCU and RNNSMUC.

An efficient hybrid neural network model in renewable energy systems

Abstract: This paper presents a hybrid neural network approach to predict wind speed automatically in renewable energy systems. Wind energy is one of the renewable energy systems with lowest cost of production of electricity with largest resources available. By the reason of the fluctuation and volatility in wind, the wind speed prediction provides the challenges in the stability of renewable energy system. The aim is to compute predicted wind speed based on hybrid model which integrates a Self Organizing Map (SOM) and Radial basis Function (RBF) neural network. The simulation result shows that the proposed approach provides better result of wind speed prediction with less error rates.

Development of genetic algorithm based hybrid neural network model for predicting the ultimate flexural strength of ferrocement elements

Abstract: This paper demonstrates the applicability of Hybrid Neural Networks that combine simple back propagation networks (BPN) and Genetic Algorithms (GAs) for predicting the ultimate flexural strength of ferrocement elements. A hybrid neural network model has been developed to predict the ultimate flexural strength of ferrocement elements with span to depth ratios (3, 6, 9 & 11.45), number of mesh layers (0, 1, 3 & 5) and percentage replacement of silicafume (0, 5, 10, 15, 20 & 25) as input parameters. The network has been trained with experimental data obtained from experimental work. The hybrid neural network model learned the relationship for predicting the ultimate flexural strength in just 300 training epochs. After successful learning, the model predicted the ultimate flexural strength satisfying all the constraints with an accuracy of 95%. The

A hybrid neuro-fuzzy simulation approach for improvement of natural gas price forecasting in industrial sectors with vague indicators

Abstract: This study presents a hybrid neural network fuzzy mathematical programming approach for improvement of natural gas price estimation in industrial sector. It is composed of artificial neural network (ANN), fuzzy linear regression (FLR), and conventional regression (CR). The preferred FLR, ANN, and CR models are selected via mean absolute percentage of error. The intelligent approach of this study is then applied to estimate natural gas price in industrial sector. Domestic sector is also used to further show the flexibility and applicability of the hybrid approach. The economic indicators used in this paper are consumer price index, population, gross domestic and annual natural gas consumption. The stated indicators could be contaminated with noise and vagueness. Moreover, there is a need to develop a hybrid approach to deal with both noise and vagueness. The input data were divided into train and test datasets. A complete sensitivity analysis has been performed by changing train and test data to show the superiority of the proposed approach. The superiority of ANN for the domestic sector and FLR for the industrial sector was proved by error analysis. The results showed that different models may be selected as preferred model, in different cases and situations. The proposed approach of this study would help policy makers to effectively manage natural gas price in vague, noisy, and complex manufacturing sectors. This is the first study that presents a hybrid approach for estimating the natural gas price in industrial sector with possible noise, non-linearity, and uncertainty.

Training a Feed-forward Neural Network with Artificial Bee Colony Based Backpropagation Method

Abstract: Back-propagation algorithm is one of the most widely used and popular techniques to optimize the feed forward neural network training. Nature inspired meta-heuristic algorithms also provide derivative-free solution to optimize complex problem. Artificial bee colony algorithm is a nature inspired meta-heuristic algorithm, mimicking the foraging or food source searching behaviour of bees in a bee colony and this algorithm is implemented in several applications for an improved optimized outcome. The proposed method in this paper includes an improved artificial bee colony algorithm based back-propagation neural network training method for fast and improved convergence rate of the hybrid neural network learning method. The result is analysed with the genetic algorithm based back-propagation method, and it is another hybridized procedure of its kind. Analysis is performed over standard data sets, reflecting the light of efficiency of proposed method in terms of convergence speed and rate.

A Hybrid Neural Network Model Predictive Control with Zone Penalty Weights for Type 1 Diabetes Mellitus

Abstract: In this paper, a hybrid neural network model is developed to predict and control the blood glucose (BG) of the patient who has type 1 diabetes mellitus (T1DM). The proposed model consists of two parts: a linear finite impulse response (FIR) model and a nonlinear autoregressive exogenous input (NARX) network. A recently developed and well-acknowledged meal simulation model of the glucose-insulin system for T1DM is employed to create virtual subjects. Data from virtual subjects are used to identify an intermediate physiological model, and then our proposed hybrid model is trained and validated based on this intermediate model. The key features of the resulting hybrid model are that it reveals satisfactory accuracy of long-term prediction and does not require an immeasurable state for model initialization. The developed hybrid model is then embedded in a nonlinear model predictive control (MPC) controller with zone penalty weights, and this closed-loop controller is implemented on these virtual subjects for ...

An Neural Network Ensemble approach based on PSO algorithm and LLE for Typhoon Intensity

Abstract: In this paper, a novel neural network ensemble forecast model is developed where the stepwise regression method are chosen for forecast factors best correlated with the series of typhoon intensity, and the main information is extracted from remaining forecast factors where Locally Linear Embedding (LLE) method is used Further the problem that network structure determination and network easily into a local solution is considered, a hybrid neural network learning Algorithm is proposed which is based on particle swarm optimization (PSO), Locally Linear Embedding and back propagation algorithm Finally, the typhoon intensity prediction experiment was conducted in the northwest Pacific Ocean from May to October 2001-2010 The results show that the mean absolute prediction error of neural network ensemble forecast model significantly less than stepwise regression method under the same conditions

A Hybrid Radial Basis Function and Particle Swarm Optimization Neural Network Approach in Forecasting the EUR/GBP Exchange Rates Returns

Abstract: The motivation for this paper is to introduce in Finance a hybrid Neural Network architecture of Adaptive Particle Swarm Optimization and Radial Basis Function (ARBF-PSO) and a Neural Network fitness function for financial forecasting purposes. This is done by benchmarking the ARBF-PSO results with those of three different Neural Networks architectures and three statistical/technical models. As it turns out, the ARBF-PSO architecture outperforms all other models in terms of statistical accuracy and trading efficiency in the examined forecasting task.

Anomaly Detection through NN Hybrid Learning with Data Transformation Analysis

Abstract: Intrusion detection system is a vital part of computer security system commonly used for precaution and detection.It is built for classifier or descriptive or predictive model to proficient classification of normal behavior from abnormal behavior of IP packets. This paper presents the solution regarding proper data transformation methods handling and importance of data analysis of complete data set which is apply on hybrid neural network approaches for used to cluster and classify normal and abnormal behavior to improve the accuracy of network based anomaly detection classifier. Because neural network classes only require the numerical form of data but IP connections or packets of network have some symbolic features which are difficult to handle without the proper data transformation analysis. For this reason, it got non redundant new NSL KDD CUP data set. The experimental results show that indicator variable is more effective as compared to the both con- ditional probabilities and arbitrary assignment method from measurement of accuracy and balance error rate.

Head-centric disparity and epipolar geometry estimation from a population of binocular energy neurons.

Abstract: We present a hybrid neural network architecture that supports the estimation of binocular disparity in a cyclopean, head-centric coordinate system without explicitly establishing retinal correspondences. Instead the responses of binocular energy neurons are gain-modulated by oculomotor signals. The network can handle the full six degrees of freedom of binocular gaze and operates directly on image pairs of possibly varying contrast. Furthermore, we show that in the absence of an oculomotor signal the same architecture is capable of estimating the epipolar geometry directly from the population response. The increased complexity of the scenarios considered in this work provides an important step towards the application of computational models centered on gain modulation mechanisms in real-world robotic applications. The proposed network is shown to outperform a standard computer vision technique on a disparity estimation task involving real-world stereo images.

Wind driven generator stability control method based on hybrid neural network

Abstract: The invention relates to a wind driven generator stability control method based on a hybrid neural network, which comprises the following steps of: (1) collecting the wind speed and corresponding active power; (2) preprocessing the filtering wave; (3) debugging the GA-BP (Genetic Algorithm - Back Propagation) algorithm; (4) training the neural network; (5) calculating the result on the display screen. The method has the advantages that the accuracy is high so that the performance of the wind driven generating system can be well fitted; the computing speed is high so that the system meets the requirement of real time property; and the realization of the GA-BP neural network algorithm through programming by using a DSP (Digital Signal Processor) is effective to increase the speed of the GA-BP neural network algorithm and better exert the parallelism of the GA-BP neural network algorithm.

Analysis of Hybrid Neural Network for Improved Performance

Abstract: this paper we take a close look at the Hybrid Neural Network Model. Hybrid model is attained by combining two Artificial Neural Networks (ANNs). In which the first model is used to perform the feature extraction task and the second one performs prediction task. This paper explores the classifying ability of the proposed hybrid model. We analyze the performance of the model, which is a compound characteristic, of which the prediction accuracy is the most important component. If the prediction accuracy of the model can be increased it will result into enhanced performance of the model. The model that has been built is under the umbrella of pattern recognition and incorporates some of the data mining techniques. Kernel Principal Component Analysis (KPCA) has been implemented in the pre-processing stage for easier subsequent analysis. By the end of the paper, the key factors that enhance the accuracy of the model have been identified and their role explained. It also has been shown that single ANN model's performance deteriorates on an unseen problem much more as compared to the hybrid model. The aim is to provide a model having better performance and accuracy. The paper focuses on the real world applications of the model.

Hybrid neural network–particle swarm method to predict global radiation over the Norte Chico (Chile)

Abstract: Solar energy estimation procedures are very important in the renewable energy field for development of mathematical models, optimization, and advanced control of processes. Solar radiation data provide information on how much of the sun’s energy strikes a surface at a location on earth during a particular time period. These data are needed for effective research into solar-energy utilization. Due to the cost and difficulty in measurement, these data are not readily available. Therefore, there is the need to develop alternative ways of generating these data. In this study, an artificial neural network (ANN) was used for the estimation of daily global solar radiation (RG) over the Norte Chico using 17 552 data measured from 21 meteorological stations (years 2004–2010) located in the south area of the Atacama Desert. The ANN was developed with particle swarm optimization. Six input parameters were used to train the network. These parameters were elevation, longitude, latitude, air temperature, relative humid...

Software of Information-Measurement System for Standardless Diagnostic of Composite Materials

Abstract: This article is devoted to software realization via NI Lab VIEW 2011 for system of the standardless diagnostic of technical objects. The solution requires methods that are fast and efficient in diagnostics, adapted for usage condition changes, oriented for wide set of objects under control and without changes in the main software structure. The structure and main modules of developed software is represented in the article. Developed software advantages are in its architecture flexib ility, high performance and reliability of data signal processing, human-engineered interface. The software of standardless diagnostics system is based on neural network classifier wh ich provides flexib le and stable knowledge base about possible classes of defects, performs effective operations with high d imensional data vectors, adapts its architecture for solving new tasks and provides the high reliab ility of control. The classifier based on hybrid neural network, ART-2 and Fu zzy-A RT neural networks for classification of defects in honeycomb panels were introduced and investigated. Described classifier during the training can automatically change its settings, reaching the highest reliab ility of the control, detect and classify subsurface defects in honeycomb panels with high reliability and accuracy, as well as defects that are located on the back side of the cladding with plottage larger than 2 cm2 and thickness of composite panel equal to 12.8 mm. The reliability of the nondestructive testing via specified classifier is more than 95%. Results of the developed special software practical usage for honeycomb panels' technical state classification were represented

Analysis of a hybrid neural network as underlying mechanism for a situation prediction engine

Abstract: This paper presents the results regarding a technique that can be used as an underlying mechanism for situation prediction. We analysed a hybrid neural network called Multi-output Adaptive Neural Fuzzy Inference System (MANFIS) and compared its predictive ability with a Multi-Layer Perceptron (MLP). The results demonstrate that, depending on the application, the use of neural networks can be considered to be a good approach for situation prediction, when combined with other techniques. Key words: situation, context, prediction, neural networks, MANFIS.

A hybrid neural network model for sea ice thickness forecasting

Abstract: Sea ice thickness is an important environment load parameter for reliability assessment and life extension decision of existing ageing offshore platforms in ice region. It is a key factor to provide accurate sea ice thickness prediction. Introducing chaos random sequence and immune mechanism into traditional genetic evolution process, Chaos immune genetic optimization algorithm is constructed. Combining the chaos immune genetic optimization algorithm and the BP neural network, a hybrid neural network models is established. The ice thickness in Bohai Sea is predicted using the hybrid neural network model, and a good fitness is revealed between the prediction and practical values. The results show that the hybrid neural network model is feasible and effective. The parameters of Weibull distribution function for ice thickness is estimated, using predicted ice thickness specimens. The assessment load in later service period is updated and more reliable environment load parameters can be provided for ageing platform assessment.