Wavelet Neural Network for Vector Prediction to Fill-In Missing Image Blocks in Wireless Transmission

Enhancing multilayer perceptron neural network using archive-based harris hawks optimizer to predict gold prices

Efficient performance of neural networks for nonlinearity error modeling of three-longitudinal-mode interferometer in nano-metrology system

In this paper, a focused time lagged recurrent neural network (FTLRNN) with gamma memory is developed to learn the dynamics of a typical liquid saturated steam heat exchanger process. This highly nonlinear process has been a significant benchmark for non-linear control design purposes, since it is characterized by non-minimum phase behaviour. It appears from the literature review that an optimal neural network (NN) model for the identification of such a highly nonlinear complex dynamical system is not currently available. This paper compares the performance of two NN configurations, namely a well-known Multi Layer Perceptron (MLP) NN model and the proposed FTLRNN model. A standard static backpropagation algorithm with momentum term has been used for both the models. It is shown that the estimated dynamic NN based model comprising of a gamma memory filter followed by a MLP based NN clearly outperforms the static MLP NN in various performance metrics such as mean square error (MSE), normalized MSE and correlation coefficients on the testing datasets. In addition, the output of the proposed NN model closely follows the desired output of the exchanger process for the testing instances. This also means that the most of the information about the rich nonlinear dynamics of the system has been extracted successfully from the training dataset and that the proposed model approximates the given system with reasonable accuracy. It is shown that the suggested dynamic NN model has a remarkable system identification capability for the problem considered in this paper. Dynamic NN model has clearly outperformed the static NN models in respect of the performance measures. The major contribution of this paper is that the FTLRNNs can elegantly be used to learn underlying highly nonlinear dynamics of the system.

Neural networks are mathematical and computer models to power nonlinear data that play a very important role in various scientific fields. They are specially used for automatic resolution of environmental problems. This study focuses on the prediction of nitrate (NO 3 - ) and sulfate (SO 4 2- ) of lake water in the Moroccan Middle Atlas. Ifrah, Iffer and Afourgagh are taken as case studies by using a number of parameters physic-chemical of water. Two methods were used: Multiple Linear Regression (MLR) and Artificial Neural Networks (ANN) Multilayer Perceptron Model (MLP). In order to choose the best neural network architecture, several statistical tests were used in conjunction with some robustness tests: Mean Square Error (MSE), mean absolute error (MAE) and correlation coefficient (R). The results showed that the models established by artificial neural network Multilayer Perceptron type (ANN-MLP) of configuration [17-8-2] are more efficient compared to those determined by the conventional method based on multiple linear regression. This performance demonstrates the existence of a nonlinear relationship between the physic-chemical characteristics of both nitrates and sulfates in the lakes waters studied that are under investigation.

Power load forecasting is of great significance to ensure the smooth operation of smart grid. Because the load generation and consumption are related to the grid internal and environmental factors external, reliable and accurate power load forecasting is undoubtedly challenging in smart grid. Since weather factors are always the leading causes that affecting power generation load in smart grid, especially in distributed photovoltaic power generation, we propose a load forecasting method to realize the forecast of the generated load under different weather conditions in this paper. We firstly investigates the combined effect of various weather factors on power load comprehensively. Specially, the parametric regression models are utilized to analyse the relationship between the power load and weather factors. Secondly, a hybrid forecasting method based on Multilayer Perceptron (MLP) neural network is proposed to achieve reliable and accurate power load forecasting of various weather conditions. Different from the existing works, we not only take into account the weather factors, but also select corresponding parametric models integrated as the additional input of the MLP neural network to predict the power load. More importantly, a modified extreme learning machine (ELM) based hierarchical learning algorithm is introduced to train the formulated model. As a result, the training process of the neutral network can be accelerated in the sense that iteration times are reduced, in which case the learning accuracy can also be guaranteed. The proposed method is evaluated on the real dataset which consist of meteorological factors and corresponding load data. The results show the proposed method outperforms the existing algorithms in prediction accuracy. The prediction error Mean Square Error(MSE) and Root Mean Squared Error(RMSE) can be reduced by 36.28% and 20.18% respectively, which ensure the reliability of the power load forecasting.

In this paper, we introduce a new intelligent combination method based on Multilayer Perceptron Neural Network (MLP‐NN) and Hybrid Genetic Algorithm (HGA) for automotive price forecasting. The combination of MLPNN and HGA lead us to accelerate convergence to the optimal weights and improve the forecasting performance. In this structure, the Levenberg‐ Marquardt (LM) algorithm is employed for training of the network, and the hybridization of Genetic Algorithm (GA) with some local search optimization techniques such as steepest descent (SD) method and quasi‐Newton methods with DFP and BFGS formula is used to perform HGA. We apply our new hybrid model to forecast the automotive prices in Iran Khodro Company which is the biggest automotive manufacturing in IRAN. Simulation results show the enough reduction in the processing iterations and forecasting error which is mean square error. The results are well promising compared to the cases when we apply MLP‐NN or hybridization of MLP‐NN and GA, individually.

In this paper, we have proposed an efficient and effective nonlinear feature domain noise suppression algorithm, motivated by the minimum mean square error (MMSE) optimization criterion. A Multi Layer Perceptron (MLP) neural network in the log spectral domain has been employed to minimize the difference between noisy and clean speech. By using this method, as a pre-processing stage of a speech recognition system, the recognition rate in noisy environments has been improved. We extended the application of the system to different environments with different noises without retraining HMM model. We trained the feature extraction stage with a small portion of noisy data which was created by artificially adding different types of noises from the NOISEX-92 database to the TIMIT speech database. In real environment, where our speech recognition systems must work, different types of noises with various SNRs exist. Our proposed method suggests four strategies based on the system capability to identify the noise type and SNR. Experimental results show that the proposed method achieves significant improvement in recognition rates.

We propose an efficient and effective nonlinear feature domain noise suppression algorithm, motivated by the minimum mean square error (MMSE) optimization criterion. Multi Layer Perceptron (MLP) neural network in the log spectral domain minimizes the difference between noisy and clean speech. By using this method as a pre-processing stage of a speech recognition system, the recognition rate in noisy environments is improved. We can extend the application of the system to different environments with different noises without re-training it. We need only to train the preprocessing stage with a small portion of noisy data which is created by artificially adding different types of noises from the NOISEX-92 database to the TIMIT speech database. Experimental results show that the proposed method can achieve significant improvement of recognition rates.

Performance analysis of all-optical logical gate using artificial neural network

Prediction of roadheaders' performance using artificial neural network approaches (MLP and KOSFM)

In this paper, field programmable gate array (FPGA) implementation of two power amplifier (PA) dynamic behavioral modeling approaches with real-valued time-delay neural network (RVTDNN) and real-valued recurrent neural network (RVRNN) architectures are presented. The proposed PA models are based on the multilayer perceptron (MLP) neural networks with delayed inputs to take into account nonlinearity and memory effects of the PA. The synoptic weights of these neural networks are dynamically updated in order to prevent any eventual change in the PA's characteristics. Both architectures have been optimized to include only six hidden neurons and implemented on FPGA using Xilinx system generator. The FPGA is preferred to the digital signal processor (DSP) because it allows parallel computation tasks and software like flexibility. The modeling performances of these architectures are compared using 16-QAM modulated test signal. The mean square error (MSE) between the desired and the actual outputs of these models are also compared.

Face recognition is one of the most efficient applications of computer authentication and pattern recognition. Therefore it attracts significant attention of researchers. In the past decades, many feature extraction algorithms have been proposed. In this paper Gabor features and Zernike moment were used to extract features from human face images for recognition application. This paper is a study for new constructive training algorithm for Multi Layer Perceptron (MLP) which is applied to face recognition application. An incremental training procedure was employed where the training patterns are learned incrementally. This algorithm started with a single training pattern and a single hidden-layer using one neuron. During neural network training, the hidden neuron is increased when the Mean Square Error (MSE) of the Training Data (TD) is not reduced or the algorithm gets stuck in a local minimum. Input patterns are trained incrementally (one by one) until all patterns of TD are selected and trained. Face recognition system structure based on a MLP neural network was constructed and was tested for face recognition. The proposed approach was tested on the UMIST database. Experimental results indicate that we can obtain an optimal architecture of neural network classifier (with the least possible number of hidden neuron) using our present constructive algorithm, and prove the effectiveness of the proposed method compared to the MLP architecture with back-propagation algorithm.

Estimating impervious surfaces from medium spatial resolution imagery using the self-organizing map and multi-layer perceptron neural networks

The precise position of the mobile station is critical for the ever increasing number of applications based on location. We introduce a novel positioning technique for positioning a GSM mobile phone in real-time. This technique is based on the GSM mobile phone feature that it can measure the signal strengths from a number of nearby base stations. We use the GSM signal strengths measured in a real environment to train an artificial neural network. The neural network is trained using the second order learning algorithm (extended Kalman filter) because of its superiority in learning speed and mapping accuracy. The mobile position can be determined with good accuracy by providing the current signal strength data to a previously trained neural network. The EKF shows its superiority to back propagation (BP) in both the general feed forward (GFF) and the multilayer perceptron (MLP) neural network architectures. The good accuracy of the calculated position with EKF training in either a GFF or MLP neural network is shown.

Machine vision researchers are working on the area of recognition of handwritten or printed text from scanned images for the purpose of digitizing documents and for reducing the errorless data entry cost. The classic difficulty of being able to correctly recognize language symbols is the complexity and the irregularity among the pictorial representation of characters due to variation in writing styles, size of symbols etc. Character recognition process depends on, how the input data is given to the system. Input data may be categorized as Online data or Offline data. Both the forms of data input have their own issues. In this paper, we are focusing on the Offline Gurmukhi character recognition from text image. There are lot of complexities associated with Gurmukhi Script. In this paper, we present a technique based on Multi Layer Perceptron (MLP) Neural Network model. Here we consider isolated handwritten Gurmukhi characters for recognition. MLP is used because it uses generalized delta learning rules and easily gets trained in less number of iterations. The proposed method in this paper detect graphical symbols by identifying lines and characters from the image. After that it analyzes the symbols by training the network using feed forward topology for a set of desired unicode characters. We achieve the performance rate of proposed system maximum up to 98.96% for recognition of symbols by using MLP neural network.