Neural Computing and Applications

Naz, Saeeda; Umar, Arif; Ahmad, Riaz; Ahmed, Saad; Shirazi, Syed; Razzak, Muhammad

doi: 10.1007/s00521-015-2051-4pmid: N/A

Character recognition for cursive script like Arabic, handwritten English and French is a challenging task which becomes more complicated for Urdu Nasta’liq text due to complexity of this script over Arabic. Recurrent neural network (RNN) has proved excellent performance for English, French as well as cursive Arabic script due to sequence learning property. Most of the recent approaches perform segmentation-based character recognition, whereas, due to the complexity of the Nasta’liq script, segmentation error is quite high as compared to Arabic Naskh script. RNN has provided promising results in such scenarios. In this paper, we achieved high accuracy for Urdu Nasta’liq using statistical features and multi-dimensional long short-term memory. We present a robust feature extraction approach that extracts feature based on right-to-left sliding window. Results showed that selected features significantly reduce the label error. For evaluation purposes, we have used Urdu printed text images dataset and compared the proposed approach with the recent work. The system provided 94.97 % recognition accuracy for unconstrained printed Nasta’liq text lines and outperforms the state-of-the-art results.

Zhu, Yulian; Xue, Jing

doi: 10.1007/s00521-015-2052-3pmid: N/A

In this paper, we propose a novel method, called random subspace method (RSM) based on tensor (Tensor-RS), for face recognition. Different from the traditional RSM which treats each pixel (or feature) of the face image as a sampling unit, thus ignores the spatial information within the face image, the proposed Tensor-RS regards each small image region as a sampling unit and obtains spatial information within small image regions by using reshaping image and executing tensor-based feature extraction method. More specifically, an original whole face image is first partitioned into some sub-images to improve the robustness to facial variations, and then each sub-image is reshaped into a new matrix whose each row corresponds to a vectorized small sub-image region. After that, based on these rearranged newly formed matrices, an incomplete random sampling by row vectors rather than by features (or feature projections) is applied. Finally, tensor subspace method, which can effectively extract the spatial information within the same row (or column) vector, is used to extract useful features. Extensive experiments on four standard face databases (AR, Yale, Extended Yale B and CMU PIE) demonstrate that the proposed Tensor-RS method significantly outperforms state-of-the-art methods.

Lin, Chih-Hong

doi: 10.1007/s00521-015-2053-2pmid: N/A

A V-belt continuously variable transmission system driven by a permanent magnet synchronous motor has much unknown nonlinear and time-varying characteristics. In order to capture the system’s nonlinear and dynamic behavior, a hybrid recurrent Laguerre-orthogonal-polynomials neural network (NN) control system with modified particle swarm optimization (PSO) is proposed for achieving online better learning capacity and faster convergence to enhance system robustness. The hybrid recurrent Laguerre-orthogonal-polynomials NN control system can perform inspected control, recurrent Laguerre-orthogonal-polynomials NN control, which involves an adaptive law, and recouped control, which involves an estimated law. Moreover, the adaptive law of online parameters in the recurrent Laguerre-orthogonal-polynomials NN is derived by means of Lyapunov stability theorem. Furthermore, two optimal learning rates of the online parameters in the recurrent Laguerre-orthogonal-polynomials NN by means of modified PSO are applied to achieve online better learning capacity and faster convergence. Finally, to show the effectiveness of the proposed control scheme, comparative studies are demonstrated by experimental results.

Kurnaz, Çetin; Engiz, Begüm; Esenalp, Murat

doi: 10.1007/s00521-015-2054-1pmid: N/A

In order to improve support for higher data rates, third-generation partnership project (3GPP) introduced dual-carrier high-speed downlink packet access (DC-HSDPA), which reaches up to 42-Mbps throughput with the use of two adjacent 5-MHz carriers in Release-8. Defining the dependence of throughput on prevailing channel parameters is crucial because a frequency-selective channel limits achieving these data rates. For this reason, DC-HSDPA throughput real field measurements were taken in different propagation environments by using the “TEMS Investigation” program. The evaluation of the measurements showed that one-parameter linear mapping methods, such as signal-to-interference ratio and channel quality indicator, are insufficient for characterizing user throughput. Therefore, this study will propose a novel mapping method with more than one variable. Although multiple linear regression gives a better normalized root-mean-square error, results have shown that frequently used artificial neural network-based mapping methods—such as those for adaptive network-based fuzzy inference system, multilayer perceptron, and generalized regression neural network (GRNN)—yield improved accuracy. From among these, user throughput can be best estimated with the use of GRNN for a commercial DC-HSDPA system, with approximately 93.3 % precision. The GRNN structure allows system designers to update system parameters to maximize user throughput.

Patil, Amit; Deka, Paresh

doi: 10.1007/s00521-015-2055-0pmid: N/A

This paper evaluates the ability of wavelet transform in improving the accuracy of artificial neural network (ANN) and adaptive neuro-fuzzy interface systems (ANFIS) models. In this study, the performance of hybrid Wavelet-ANN and Wavelet-ANFIS models for estimating daily evapotranspiration in arid regions was evaluated. Prior to the development of models, gamma test was used to identify the best input combinations that could be used under limited data scenario. Performance of the proposed hybrid models was compared to ANN, ANFIS, and conventionally used Hargreaves equation. The results revealed that use of wavelet transform as data preprocessing technique enhanced the efficiency of ANN and ANFIS models. Wavelet-ANN and Wavelet-ANFIS performed reasonably better than other models. Better handling of wavelet-decomposed input variables enabled Wavelet-ANN models to perform slightly better than the Wavelet-ANFIS models. W-ANN2 (RMSE = 0.632 mm/day and R = 0.96) was found to be the best model for estimating daily evapotranspiration in arid regions. The proposed W-ANN2 model used second-level db3 wavelet-decomposed subseries of temperature and previous day evapotranspiration values as inputs. The study concludes that hybrid Wavelet-ANN and Wavelet-ANFIS models can be effectively used for modeling evapotranspiration.

Strnad, Damjan; Nerat, Andrej; Kohek, Štefan

doi: 10.1007/s00521-015-2056-zpmid: N/A

Soccer match attendance is an example of group behavior with noisy context that can only be approximated by a limited set of quantifiable factors. However, match attendance is representative of a wider spectrum of context-based behaviors for which only the aggregate effect of otherwise individual decisions is observable. Modeling of such behaviors is desirable from the perspective of economics, psychology, and other social studies with prospective use in simulators, games, product planning, and advertising. In this paper, we evaluate the efficiency of different neural network architectures as models of context in attendance behavior by comparing the achieved prediction accuracy of a multilayer perceptron (MLP), an Elman recurrent neural network (RNN), a time-lagged feedforward neural network (TLFN), and a radial basis function network (RBFN) against a multiple linear regression model, an autoregressive moving average model with exogenous inputs, and a naive cumulative mean model. We show that the MLP, TLFN, and RNN are superior to the RBFN and achieve comparable prediction accuracy on datasets of three teams from the English Football League Championship, which indicates weak importance of context transition modeled by the TLFN and the RNN. The experiments demonstrate that all neural network models outperform linear predictors by a significant margin. We show that neural models built on individual datasets achieve better performance than a generalized neural model constructed from pooled data. We analyze the input parameter influences extracted from trained networks and show that there is an agreement between nonlinear and linear measures about the most significant attributes.

Rezaei, Hossein; Rahmati, Mahmoud; Modarress, Hamid

doi: 10.1007/s00521-015-2057-ypmid: N/A

In this work, cationic (Mg2+, Ca2+, Sr2+, and Ba2+) substitution in X and Y faujasite zeolite structures and their effects on capacity of zeolites for methane adsorption were studied by applying multiple linear regression and expert adaptive neuro-fuzzy inference system (ANFIS) . Temperature, pressure, and molecular weight of cations were used as the input parameters. The results obtained from application of the proposed ANFIS model showed that at high pressures, the zeolite with smaller cation in their structure have higher methane adsorption capacity. The root-mean-square error, square correlation coefficient (R 2), mean absolute error, and percentage of mean absolute relative error for X and Y faujasite zeolites were evaluated, which indicated that ANFIS model can accurately predict the adsorption of methane gas on X and Y zeolites in the presence of the substituted cations.

Karabulut, Şener; Karakoç, Halil

doi: 10.1007/s00521-015-2058-xpmid: N/A

In the present study, aluminum alloy 7075 (Al7075)-based open-cell silicon carbide (SiC) foam composite was fabricated and the machinability of both Al7075 and the open-cell SiC foam Al metal matrix composite was investigated during milling using an uncoated carbide tool. The machining trials were conducted using the Taguchi L27 full-factorial orthogonal array, and the milling parameters were optimized for surface roughness. Analysis of variance was employed to determine the effect of the cutting variables on surface roughness. The experimental results were evaluated by signal-to-noise ratio, 3D surface graphs, artificial neural networks (ANNs) and main effect graphs. The analysis results show that the feed rate was the most significant milling parameter affecting surface roughness of both Al7075 and the open-cell SiC foam composite. Prediction models have been developed for the surface roughness through regression analysis and ANNs. Confirmation experiments were performed to identify the performance of mathematical models, and the surface roughness was predicted with a mean squared error equal to 1.6 and 0.24 % in the milling of Al7075 and open-cell SiC foam composite, respectively. The test result showed that the three-dimensional open-pore SiC foam network reinforcement was restricted the movement of the soft matrix and provided an acceptable surface quality in the milling of MMCs.

Arunkumar, R.; Karthigaikumar, P.

doi: 10.1007/s00521-015-2059-9pmid: N/A

This paper presents the retina-based disease diagnosis through deep learning-based feature extraction method. This process helps in developing automated screening system, which is capable of diagnosing retina for diseases such as age-related molecular degeneration, diabetic retinopathy, macular bunker, retinoblastoma, retinal detachment, and retinitis pigmentosa. Some of these diseases share a common characteristic, which makes the classification difficult. In order to overcome the above-mentioned problem, deep learning feature extraction and a multi-class SVM classifier are used. The main contribution of this work is the reducing the dimension of the features required to classify the retinal disease, which enhances the process of reducing the system requirement as well as good performance.

Choi, Sung-Uk; Choi, Byungwoong; Lee, Seonmin

doi: 10.1007/s00521-015-2062-1pmid: N/A

Local scour around bridge piers is a complicated physical process and involves highly three-dimensional flows. Thus, the scour depth, which is directly related to the safety of a bridge, cannot be given in the form of the exact relationship of dependent variables via an analytical method. This paper proposes the use of the adaptive neuro-fuzzy inference system (ANFIS) method for predicting the scour depth around a bridge pier. Five variables including mean velocity, flow depth, size of sediment particles, critical velocity for particles’ initiation of motion, and pier width were used for the scour depth. For comparison, predictions by the artificial neural network (ANN) model were also provided. Both the ANN model and ANFIS method were trained and validated. The findings indicate that the modeling with dimensional variables yields better predictions than when normalized variables are used. The ANN model was applied to a field-scale dataset. Prediction results indicated that the errors are much larger compared to the case of a laboratory-scale dataset. The MAPE by the ANN model trained with part of the field data was not seriously different from that by the model trained with the laboratory data. However, the application of the ANFIS method improved the predictions significantly, reducing the MAPE to the half of that by the ANN model. Five selected empirical formulas were also applied to the same dataset, and Sheppard and Melville’s formula was found to provide the best prediction. However, the MAPEs for the scour depths predicted by empirical formulas are much larger than MAPEs by either the ANN or the ANFIS method. The ANFIS method predicts much better if the range of the training dataset is sufficiently wide to cover the range of the application dataset.