This paper deals with artificial neural network modeling of diesel engine fueled with jatropha oil to predict the unburned hydrocarbons, smoke, and NOx emissions. The experimental data from the literature have been us...
详细信息
This paper deals with artificial neural network modeling of diesel engine fueled with jatropha oil to predict the unburned hydrocarbons, smoke, and NOx emissions. The experimental data from the literature have been used as the data base for the proposed neural network model development. For training the networks, the injection timing, injector opening pressure, plunger diameter, and engine load are used as the input layer. The outputs are hydrocarbons, smoke, and NOx emissions. The feed forward back propagation learning algorithms with two hidden layers are used in the networks. For each output a different network is developed with required topology. The artificial neural network models for hydrocarbons, smoke, and NOx emissions gave R-2 values of 0.9976, 0.9976, and 0.9984 and mean percent errors of smaller than 2.7603, 4.9524, and 3.1136, respectively, for training data sets, while the R-2 values of 0.9904, 0.9904, and 0.9942, and mean percent errors of smaller than 6.5557, 6.1072, and 4.4682, respectively, for testing data sets. The best linear fit of regression to the artificial neural network models of hydrocarbons, smoke, and NOx emissions gave the correlation coefficient values of 0.98, 0.995, and 0.997, respectively.
Reducing the level of the targets corresponding to training samples for a machine classifier using the outputs of an auxiliary classifier is interesting because it allows to save expressive power unnecessarily dedicat...
详细信息
Reducing the level of the targets corresponding to training samples for a machine classifier using the outputs of an auxiliary classifier is interesting because it allows to save expressive power unnecessarily dedicated to increase the output level of well-classified samples. In this paper we propose an iterative form of this selective reduction of target levels with a simple linear reduction schedule. Extensive simulations show that the proposed method has not only a performance better than or equal to conventional training or using static versions of the reduction, but also with respect to support vector machines (SVM). This potential advantage is accompanied by a smaller size and a design effort not much higher than the corresponding SVM, thus making the proposed method very attractive for practical applications. (C) 2009 Elsevier B.V. All rights reserved.
Most current Artificial Neural Network (ANN) models are based on highly simplified brain dynamics. They have been used as powerful computational tools to solve complex pattern recognition, function estimation, and cla...
详细信息
Most current Artificial Neural Network (ANN) models are based on highly simplified brain dynamics. They have been used as powerful computational tools to solve complex pattern recognition, function estimation, and classification problems. ANNs have been evolving towards more powerful and more biologically realistic models. In the past decade, Spiking Neural Networks (SNNs) have been developed which comprise of spiking neurons. Information transfer in these neurons mimics the information transfer in biological neurons, i.e., via the precise timing of spikes or a sequence of spikes. To facilitate learning in such networks, new learning algorithms based on varying degrees of biological plausibility have also been developed recently. Addition of the temporal dimension for information encoding in SNNs yields new insight into the dynamics of the human brain and could result in compact representations of large neural networks. As such, SNNs have great potential for solving complicated time-dependent pattern recognition problems because of their inherent dynamic representation. This article presents a state-of-the-art review of the development of spiking neurons and SNNs, and provides insight into their evolution as the third generation neural networks.
In order to improve solving Support Vector Machine algorithm, an improved learning algorithm of the parallel SMO is proposed. According to this algorithm, the master CPU averagely distributes primitive training set to...
详细信息
ISBN:
(纸本)9781424447053
In order to improve solving Support Vector Machine algorithm, an improved learning algorithm of the parallel SMO is proposed. According to this algorithm, the master CPU averagely distributes primitive training set to slave CPUs so that they can almost independently run serial SMO on their respective training set. As it adopts the strategies of buffer and shrink, the speed of the parallel training algorithm is increased, which is showed in the experiments of parallel SMO based on the dataset of MNIST. The experiments indicate that the parallel SMO algorithm has good performance in solving large-scale SVM.
Ensemble methods used for classification and regression have been shown that they are superior than other methods, teoritically and empirically. Adapting this method on time-series prediction is done by using boosting...
详细信息
ISBN:
(纸本)9781424449996
Ensemble methods used for classification and regression have been shown that they are superior than other methods, teoritically and empirically. Adapting this method on time-series prediction is done by using boosting algorithm. On boosting algorithm, recurrent neural networks (RNN) are generated, each for training on a different set of examples on time-series data, then the results for each of this base learners will be combined and resulting on a final hypothesis. The difference between our algorithm and the original algorithm is the introduction of a new parameter for tuning the boosting influence on given examples. Our boosting result is then tested on real time-series forecasting, using a natural dataset and function-generated time series. On the experiment result, it can be proved that ensemble method that we used is better than standard method, backpropagation through time for one step ahead time series prediction.
Considering that inputs of a process neural network (PNN) are generally time-varying functions while the inputs of many practical problems are discrete values of multiple series, in this paper, a process neural networ...
详细信息
ISBN:
(纸本)9781424427239
Considering that inputs of a process neural network (PNN) are generally time-varying functions while the inputs of many practical problems are discrete values of multiple series, in this paper, a process neural network with discrete inputs is presented to provide improved forecasting results for solving the complex time series prediction. The presented method first makes discrete input series carry out Walsh transformation, and submits the transformed series to the network for training. It can solve the problem of space-time aggregation operation of PNN. In order to examine the effectiveness of the presented method, the actual data of sunspots during 1749-2007 are employed. To predict the number of sunspots, the suitability of the developed model is examined in comparison with the other models to show its superiority and be an effective way of improving forecasting accuracy of networks.
Traditional PED neural network adopts BP learning algorithm. However, without accurate gradients, its initial MSE is too large and the procedure of convergence may be unstable. A modified PSO (MPSO) algorithm is intro...
详细信息
ISBN:
(纸本)9787810778022
Traditional PED neural network adopts BP learning algorithm. However, without accurate gradients, its initial MSE is too large and the procedure of convergence may be unstable. A modified PSO (MPSO) algorithm is introduced to training the PID neural network. The WSO algorithm does not need any gradient information. It can keep large variety all along and solve premature convergence, which is a major problem in basic PSO algorithm. Simulation results show MPSO algorithm is the best learning algorithm for PID neural network.
Forecasting pests emergence levels plays a significant role in regional crop planting and management. The accuracy, which is derived from the accuracy of the forecasting approach used, will determine the economics of ...
详细信息
ISBN:
(纸本)9781441902108
Forecasting pests emergence levels plays a significant role in regional crop planting and management. The accuracy, which is derived from the accuracy of the forecasting approach used, will determine the economics of the operation of the pests prediction. Conventional methods including time series, regression analysis or ARMA model entail exogenous input together with a number of assumptions. The use of neural networks has been shown to be a cost-effective technique. But their training, usually with back-propagation algorithm or other gradient algorithms, is featured with some drawbacks such as very slow convergence and easy entrapment in a local minimum. This paper presents a hybrid approach of neural network with particle swami optimization for developing the accuracy of predictions. The approach is applied to forecast Alternaria alternate Keissl emergence level of the WuLong Country, one of the most important tobacco planting areas in Chongqing. Traditional ARMA model and BP neural network are investigated as comparison basis. The experimental results show that the proposed approach can achieve better prediction performance.
A method for designing and training process neural networks (PNN) based on quantum genetic algorithm (QGA) was presented in this paper. Firstly, an improved quantum genetic algorithm based on Bloch coordinates of qubi...
详细信息
ISBN:
(纸本)9781846260650
A method for designing and training process neural networks (PNN) based on quantum genetic algorithm (QGA) was presented in this paper. Firstly, an improved quantum genetic algorithm based on Bloch coordinates of qubits is put forward. This method is integrated into the training of process neural networks, the number of genes on a chromosome is determined by the number of weight parameters and colony encoding is completed;the present optimal chromosome is obtained by means of chromosome assessment, and taking each qubit in this optimal chromosome as the goal, individuals in the colony are updated by new quantum rotation gate. In this method, each chromosome carries three chains of genes, therefore it can extend ergodicity for solution space and accelerate optimization process. Taking the pattern classification of two groups of two-dimensional trigonometric functions as an example, the simulation results show that the method is feasible and effective.
In this paper, a novel function series expansion method based on functional network model is proposed, and a functional network model for functions in several variables series expansion and learning algorithm are give...
详细信息
ISBN:
(纸本)9781424447541
In this paper, a novel function series expansion method based on functional network model is proposed, and a functional network model for functions in several variables series expansion and learning algorithm are given, the learning of parameters of the functional networks is carried out by the solving linear equations The simulation results show that the proposed approach is more efficient and feasible in function series expansion By this algorithm, we only need the sample space of the original functions So this algorithm has the value of application in industries
暂无评论