Multi-hidden Layers Research Articles

Machine learning based prediction for bulk porosity and static elastic modulus of Yungang Grottoes sandstone

In this work, four mainstream machine learning (ML) techniques are used to evaluate the bulk porosity and static elastic modulus of weathered Yungang Grottoes sandstone. Datasets are gathered from the experiments, which includes 432 groups effective experimental data including 8 inputs features. bulk porosity and static elastic modulus were considered as outputs to determine the weathering degrees of Yungang Grottoes sandstone. The 4 performance criteria were used to evaluate the ML models. Results demonstrate that the Artificial Neural Network (ANN) is the best-fitted models for estimating the bulk porosity and static elastic modulus compared to Multiple Linear Regression (MLR), Support Vector Regression (SVR), Gaussian Process Regression (GPR). The accuracy of the trained model for static elastic modulus is slightly higher than that of bulk porosity. The GPR and ANN model can accurately predict the bulk porosity and static elastic modulus in training stages. The ANN with multi-hidden layers developed is competent with high degree of precision and generalization ability for bulk porosity and static elastic modulus compared to other selected regression-based ML models (MLR, SVR, and GPR). The coefficient of determinations of ANN in the range of (0.9537–0.9641) during the testing stages is more stable and higher than that of (0.8883–0.9453) other built ML models. The prediction efficiency of pretrained ANN model was well adjusted for the actual and forecast datasets at the training and testing stages, and the error range was no more than 0.7% and 0.15 GPa at both stages of prediction for bulk porosity and static elastic modulus respectively. And the ANN based static elastic modulus prediction model’s error proportions significantly decreased and were confined to a modest range between + 10% and − 10%. The proposed surrogate models are valid for the bulk porosity ranging from 7 to 14% and the static elastic modulus ranging from 0.7 to 1.4 Gpa, which can be utilized for the accurate and fast prediction of the weathering degrees of Yungang Grottoes sandstone.

Q-Deformed and delta-parametrized A-generalized logistic function induced Banach space valued multivariate multi layer neural network approximations

Abstract. Here we research the multivariate quantitative approximation of Banach space valued continuous multivariate functions on a box or RN, N ∈ N, by the multivariate normalized, quasi-interpolation, Kantorovich type and quadrature type neural network operators. We investigate also the case of approximation by iterated multilayer neural network operators of the last four types. These approximations are achieved by establishing multidimensional Jackson type inequalities involving the multivariate modulus of continuity of the engaged function or its partial derivatives. Our multivariate operators are defined by using a multidimensional density function induced by a q-deformed and λ-parametrized A-generalized logistic function, which is a sigmoid function. The approximations are pointwise and uniform. The related feed-forward neural network are with one or multi hidden layers. Mathematics Subject Classification (2010): 41A17, 41A25, 41A30, 41A36. Keywords: Multi-layer approximation, q-deformed and λ-parametrized A- generalized logistic function, multivariate neural network approximation, quasi-interpolation operator, Kantorovich type operator, quadrature type operator, multivariate modulus of continuity, iterated approximation.

Improved deep neural network for predicting structural response of stiffened cylindrical shells to far-field underwater explosion

This paper is focused on the application of the machine learning for predicting structural response of ring stiffened cylindrical shells subjected to far-field underwater explosion. Three deep neural network (DNN) models are combined to predict the progressive plastic deformation behaviour of stiffened cylindrical shells subjected to the shock wave produced by a TNT charge. In the DNN regression models, the input variables include the charge mass, stand-off distance, detonation depth, cylindrical shell thickness, stiffener web and flange. These three DNN models composed of multi hidden layers are trained based on the data collection originated from a series of simulations performed by the ABAQUS finite element solver. Grid search is introduced to the hyperparameter tuning for deep learning, improving the prediction performance of machine learning methods. The DNN models are compared to the models trained using another three methods: multiple linear regression, decision tree and k-nearest neighbour. An improved Adam optimisation algorithm is used to increase the accuracy of prediction. The challenges encountered during the predictions are discussed to provide a more comprehensive insight into the advantage of the proposed DNN models. The DNN regression models can well predict the permanent plastic deformation and strain of stiffened cylindrical shells.

Q-Deformed and L-parametrized hyperbolic tangent function relied complex valued multivariate trigonometric and hyperbolic neural network approximations

Here we study the multivariate quantitative approximation of complex valued continuous functions on a box of RN , N ∈ N, by the multivariate normalized type neural network operators. We investigate also the case of approximation by iterated multilayer neural network operators. These approximations are achieved by establishing multidimen-sional Jackson type inequalities involving the multivariate moduli of continuity of the en- gaged function and its partial derivatives. Our multivariate operators are defined by using a multidimensional density function induced by a q-deformed and λ-parametrized hyper-bolic tangent function, which is a sigmoid function. The approximations are pointwise and uniform. The related feed-forward neural network are with one or multi hidden layers. The basis of our theory are the introduced multivariate Taylor formulae of trigonometric and hyperbolic type.

Statistical and Artificial Neural Network Analysis for Corrosion of Mild Steel in Hydrochloric Acid in Presence of Eco-Friendly Inhibitor

In present work, the inhibition performance of the Portulaca Grandiflora Leaf Extract (PGL), as environmental-friendly corrosion inhibitor, for low-carbon steel in 0.5 M hydrochloric acid solution at variable inhibitor concentrations and temperatures is evaluated by mass loss technique. The dependent variable was corrosion rate, while the independent variables were inhibitor concentration and temperatures. Several mathematical and artificial neural network (ANN) models have been suggested. A computer aided program is used during regression and estimation processes. Several models were used. Polynomial – individual effect, polynomial – interaction effect, linear effect, exponential growth, and piecewise regression models were estimated. Results show that the Piecewise regression model was the best one with high correlation coefficient (R2) equal to 0.9994. For ANN studies, Linear Model (LM), Radial Basis Function (RBF), and Multi-Layer Perceptron (MLP) were evaluated. The data were divided into training and testing. MLP of two inputs, multi-hidden layers, and one output (2:2-8-1:1) was the highly accurate artificial neural network model ANN with a high correlation coefficient (R2=0.9826). The effect of temperature was lower than the effect of PGL concentration as shown by mathematical and ANN analysis.

Crop Field Monitoring and Disease Detection of Plants in Smart Agriculture using Internet of Things

The Internet of Things can be defined as the network of physical objects that have sensors, software, and other technologies built into them in order to communicate and exchange data with other systems and devices over the internet. In intelligent agricultural advancements to increase the quality of agriculture, the Internet of Things (IoT) can be used. The manual monitoring of plant diseases is quite challenging. It demands enormous effort, expertise in the diseases of plants and the considerable time required for processing. The idea of automation in Smart Agriculture is implemented using the Internet of Things (IoT). They help monitor the plant leaf conditions, control water irrigation, gather images using installed IoT system which includes NodeMCU, cameras, soil moisture, temperature sensors and detect diseases in plants on the datasets collected from leaves. To detect plant diseases, image processing is applied. The detection of diseases comprises the acquisition of images, image pre-processing, segmenting an image, extracting and classifying features. In addition, the performance of two machine-learning techniques, such as a linear and polynomial kernel multi hidden extreme machine (MELM) and a support vector machine (SVM), has been studied. This paper discussed how plant diseases could be detected via images of their leaves. This analysis seeks to validate a proposed system for an ap-propriate solution to the IoT-based environmental surveillance, water irrigation system management and an efficient approach for leaf disease detection on plants. The proposed multi hidden layers extreme machine classification delivers good performance of 99.12% in the classification of leaf diseases in comparison to the Support Vector Machine classification, which gives 98%.

Fault identification of rolling bearing based on multi hidden layers wavelet convolution extreme learning neural network

Fault identification of rolling bearing based on multi hidden layers wavelet convolution extreme learning neural network

Experimental investigation and prediction of tribological behavior of unidirectional short castor oil fiber reinforced epoxy composites

The present study aims at introducing a newly developed natural fiber called castor oil fiber, termed ricinus communis, as a possible reinforcement in tribo-composites. Unidirectional short castor oil fiber reinforced epoxy resin composites of different fiber lengths with 40% volume fraction were fabricated using hand layup technique. Dry sliding wear tests were performed on a pin-on-disc tribometer based on full factorial design of experiments (DoE) at four fiber lengths (5, 10, 15, and 20 mm), three normal loads (15, 30, and 45 N), and three sliding distances (1,000, 2,000, and 3,000 m). The effect of individual parameters on the amount of wear, interfacial temperature, and coefficient of friction was studied using analysis of variance (ANOVA). The composite with 5 mm fiber length provided the best tribological properties than 10, 15, and 20 mm fiber length composites. The worn surfaces were analyzed under scanning electron microscope. Also, the tribological behavior of the composites was predicted using regression, artificial neural network (ANN)-single hidden layer, and ANN-multi hidden layer models. The confirmatory test results show the reliability of predicted models. ANN with multi hidden layers are found to predict the tribological performance accurately and then followed by ANN with single hidden layer and regression model.

Fluorescence detection of multiple kinds of pesticides with multi hidden layers neural network algorithm

Pesticide residues have become a global concern because of the threaten to human health. To realize the non-destructive detection of the pesticides containing in the drinking water, this work introduces multiple layers neural network algorithm to predict multiple kinds of pesticides with a single model of fluorescence spectra. To demonstrate the effectiveness of the algorithm, four traditional pesticides (zhongshengmycin, paclobutrazol, boscalid and pyridaben) were dissolved in the drinking water to achieve different concentrations and then the fluorescence spectra of these samples were measured. A model with four hidden layers was built using these fluorescence spectra. 50 samples were used to validate the model and the results demonstrated the feasibility of the model.

A Model Transfer Learning Framework With Back-Propagation Neural Network for Wine and Chinese Liquor Detection by Electronic Nose

Electronic nose, as a non-destructive instrument, is widely used in the field of gas analysis. In this work, E-nose was employed to distinguish wines and Chinese liquors by means of a machine learning technique. First, a multi-hidden layers Back-Propagation Neural Network (BPNN) was designed to build an identification model for the classification of different wines. Then, a BPNN-based transfer-learning framework was developed with minimal changes to the architecture of the BPNN-based model which was trained on the wine sample dataset. Experimental results revealed that the BPNN-based model performed with a 98.27% accuracy in identifying different wines, and the BPNN-based transfer-learning framework performed with a 93.4% accuracy in identifying Chinese liquors by only re-training the output layer. This reduced the model training costs compared with the complete retraining of a new classification model. Results demonstrated the effectiveness of the proposed BPNN-based transfer-learning model, which was capable of identifying different kinds of wines based on their own properties and could be easily applied to the classification of Chinese liquors. The model-based transfer learning framework offered promising potential for different classification tasks of various beverage.

DAEN: Deep Autoencoder Networks for Hyperspectral Unmixing

Spectral unmixing is a technique for remotely sensed image interpretation that expresses each (possibly mixed) pixel as a combination of pure spectral signatures (endmembers) and their fractional abundances. In this paper, we develop a new technique for unsupervised unmixing which is based on a deep autoencoder network (DAEN). Our newly developed DAEN consists of two parts. The first part of the network adopts stacked autoencoders (SAEs) to learn spectral signatures, so as to generate a good initialization for the unmixing process. In the second part of the network, a variational autoencoder (VAE) is employed to perform blind source separation, aimed at obtaining the endmember signatures and abundance fractions simultaneously. By taking advantage from the SAEs, the robustness of the proposed approach is remarkable as it can unmix data sets with outliers and low signal-to-noise ratio. Moreover, the multihidden layers of the VAE ensure the required constraints (nonnegativity and sum-to-one) when estimating the abundances. The effectiveness of the proposed method is evaluated using both synthetic and real hyperspectral data. When compared with other unmixing methods, the proposed approach demonstrates very competitive performance.

Forecasting the Volatility of Stock Market Index Using the Hybrid Models with Google Domestic Trends

In order to improve the forecasting accuracy of the volatilities of the markets, we propose the hybrid models based on artificial neural networks with multi-hidden layers in this paper. Specifically, the hybrid models are built using the estimated volatilities obtained from GARCH family models and Google domestic trends (GDTs) as input variables. We further carry out many experiments varying the number of layers and activation functions to obtain the accurate hybrid model for forecasting volatility. The proposed models are applied to forecast weekly and monthly volatilities of S&P 500 index to verify their accuracy. The performance comparison results show that the hybrid models with GDTs outperform clearly the predicted results with GARCH family models and the hybrid models without GDTs in forecasting the volatility of actual market. We also provide the experiment results with graphs to illustrate the efficiency of models.

The Rayleigh Fading Channel Prediction via Deep Learning

This paper presents a multi‐time channel prediction system based on backpropagation (BP) neural network with multi‐hidden layers, which can predict channel information effectively and benefit for massive MIMO performance, power control, and artificial noise physical layer security scheme design. Meanwhile, an early stopping strategy to avoid the overfitting of BP neural network is introduced. By comparing the predicted normalized mean square error (NMSE), the simulation results show that the performances of the proposed scheme are extremely improved. Moreover, a sparse channel sample construction method is proposed, which saves system resources effectively without weakening performances.

Action Recognition using Key-Frame Features of Depth Sequence and ELM

Recently, the rapid development of inexpensive RGB-D sensor, like Microsoft Kinect, provides adequate information for human action recognition. In this paper, a recognition algorithm is presented in which feature representation is generated by concatenating spatial features from human contour of key frames and temporal features from time difference information of a sequence. Then, an improved multi-hidden layers extreme learning machine is introduced as classifier. At last, we test our scheme on the public UTD-MHAD dataset from recognition accuracy and time consumption.

Research of stacked denoising sparse autoencoder

Learning results depend on the representation of data, so how to efficiently represent data has been a research hot spot in machine learning and artificial intelligence. With the deepening of the deep learning research, studying how to train the deep networks to express high dimensional data efficiently also has been a research frontier. In order to present data more efficiently and study how to express data through deep networks, we propose a novel stacked denoising sparse autoencoder in this paper. Firstly, we construct denoising sparse autoencoder through introducing both corrupting operation and sparsity constraint into traditional autoencoder. Then, we build stacked denoising sparse autoencoders which has multi-hidden layers by layer-wisely stacking denoising sparse autoencoders. Experiments are designed to explore the influences of corrupting operation and sparsity constraint on different datasets, using the networks with various depth and hidden units. The comparative experiments reveal that test accuracy of stacked denoising sparse autoencoder is much higher than other stacked models, no matter what dataset is used and how many layers the model has. We also find that the deeper the network is, the less activated neurons in every layer will have. More importantly, we find that the strengthening of sparsity constraint is to some extent equal to the increase in corrupted level.

Earthquakes magnitude predication using artificial neural network in northern Red Sea area

Since early ages, people tried to predicate earthquakes using simple observations such as strange or atypical animal behavior. In this paper, we study data collected from past earthquakes to give better forecasting for coming earthquakes. We propose the application of artificial intelligent predication system based on artificial neural network which can be used to predicate the magnitude of future earthquakes in northern Red Sea area including the Sinai Peninsula, the Gulf of Aqaba, and the Gulf of Suez. We present performance evaluation for different configurations and neural network structures that show prediction accuracy compared to other methods. The proposed scheme is built based on feed forward neural network model with multi-hidden layers. The model consists of four phases: data acquisition, pre-processing, feature extraction and neural network training and testing. In this study the neural network model provides higher forecast accuracy than other proposed methods. Neural network model is at least 32% better than other methods. This is due to that neural network is capable to capture non-linear relationship than statistical methods and other proposed methods.

Multi-objective simultaneous prediction of waterborne coating properties

Multi-objective simultaneous prediction of waterborne coating properties was studied by the neural network combined with programming. The conditions of network with one input layer, three hidden layers and one output layer were confirmed. The monomers mass of BA, MMA, St and pigments mass of TiO2 and CaCO3 were used as input data. Four properties, which were hardness, adhesion, impact resistance and reflectivity, were used as network output. After discussing the hidden layer neurons, learn rate and the number of hidden layers, the best net parameters were confirmed. The results of experiment show that multi-hidden layers was advantageous to improve the accuracy of multi-objective simultaneous prediction. 36 kinds of coating formulations were used as the training subset and 9 acrylate waterborne coatings were used as testing subset in order to predict the performance. The forecast error of hardness was 8.02% and reflectivity was 0.16%. Both forecast accuracy of adhesion and impact resistance were 100%.