Form Of Multilayer Perceptron Research Articles

Examining the effectiveness of artificially replicated lake systems in predicting eutrophication indicators: a comparative data-driven analysis

Abstract Data-driven models for the prediction of lake eutrophication essentially rely on water quality datasets for a longer duration. If such data are not readily available, lake management through data-driven modeling becomes impractical. So, a novel approach is presented here for the prediction of eutrophication indicators, such as dissolved oxygen, Secchi depth, total nitrogen, and total phosphorus, in the waterbodies of Assam, India. These models were developed using water quality datasets collected through laboratory investigation in artificially simulated lake systems. Two artificial prototype lakes were eutrophied in a controlled environment with the gradual application of wastewater. A periodic assessment of water quality was done for model development. Data-driven modeling in the form of multilayer perceptron (MLP), time-delay neural network (TDNN), support vector regression (SVR), and Gaussian process regression (GPR) were utilized. The trained model's accuracy was evaluated based on statistical parameters and a reasonable correlation was observed between targeted and model predicted values. Finally, the trained models were tested against some natural waterbodies in Assam and a satisfactory prediction accuracy was obtained. TDNN and GPR models were found superior compared to other methods. Results of the study indicate feasibility of the adopted modeling approach in predicting lake eutrophication when periodic water quality data are limited for the waterbody under consideration.

Short-term load forecasting using neural networks and global climate models: An application to a large-scale electrical power system

This paper focuses on the development of shallow and deep neural networks in the form of multi-layer perceptron, long-short term memory, and gated recurrent unit to model the short-term load forecasting problem. Different model architectures are tested, and global climate model information is used as input to generate more accurate forecasts. A real study case is presented for the Brazilian interconnected power system and the results generated are compared with the forecasts from the Brazilian Independent System Operator model. In general terms, results show that the bidirectional versions of long-short term memory and gated recurrent unit produce better and more reliable predictions than the other models. From the obtained results, the recurrent neural networks reach Nash-Sutcliffe values up to 0.98, and mean absolute percentile error values of 1.18%, superior than the results obtained by the Independent System Operator models (0.94 and 2.01% respectively). The better performance of the neural network models is confirmed under the Diebold-Mariano pairwise comparison test.

Efficacy of linear multiple regression and artificial neural network for long-term rainfall forecasting in Western Australia

Precipitation is one of the most intrinsic resources for manifold industrial activities all over Western Australia; consequently, immaculate rainfall prediction is indispensable for flood mitigation as well as water resources management. This study investigated the performance of artificial neural networks (ANN) and Linear multiple regression (LMR) analysis to forecast long-term seasonal spring rainfall in Western Australia, using lagged El Nino Southern Oscillation (ENSO) and Indian Ocean Dipole (IOD) as potential climatic phenomena. The ANN was developed in the form of multilayer perceptron using Levenberg–Marquardt algorithm and subsequently LMR was used with statistical significance for future spring rainfall forecast. The total climatic dataset has been divided into calibration and testing phases to determine the efficacy of the developed models. Different statistical skill tests such as root mean square error (RMSE), mean absolute error (MAE), and Willmott index of agreement ‘d’ were used to assess the efficacy of LMR and ANN modelling. In general, LMR has lower MAE and RMSE values as compared to ANN for most of the stations during calibration and testing periods, whereas ANN models performed better than LMR models based on ‘d’ values. The overall statistical analysis paradigm suggests the efficacy of LMR over ANN models for rainfall forecasting using more climatic variables. As a result, the developed LMR model, incorporated with lagged global climate indices, will facilitate the adequate preparedness for the risks associated with potential droughts in the study region.

Mixing time prediction with artificial neural network model

The study presents the methodology for artificial neural network model learning for the purpose of predicting mixing time on a set including 782 data depending on: type, diameter and position of the impeller in the vessel without baffles and the Newtonian fluid flow character and position of the tracer dosing point in the vessel. Numerous models in the form of MLP (multilayer perceptron) were trained, differing in the number of neurons in the hidden layer of the network, and the best quality MLP 15-8-1 model was determined, which can be successfully used to predict mixing time in a vessel of any configuration (mean error MAPE = 13%). Based on the predict error MAPE level, the quality of networks and the quality of the obtained predicts was assessed, whereas based on the model of cumulated risk Rsk-20 the level of likelihood of obtaining the predict of mixing time bearing the MAPE error>20% was estimated.

A Novel Approach in Determining Neural Networks Architecture to Classify Data With Large Number of Attributes

One of the challenges in the successful implementation of deep neural networks (DNN) lies on the determination of its architecture, in terms of the number of hidden layers and neurons for each hidden layer. In this research, a new approach is proposed to determine the neural networks architecture especially in the form of Multi-layer Perceptron (MLP) which will later be used as a machine learning method to classify data with large number of attribute. The new approach is proposed since the previous approaches are no longer applicable as general guidelines to determine the architecture of neural networks. Thus, the proposed approach aims to determine the number of hidden layers by using principal component analysis (PCA), while the number of neurons for each hidden layer is determined by using K-Means clustering. The determined neural network architecture is utilized to classify data with large number of attribute, such as the Gas Sensor Array Drift dataset which has 128 input attributes and six output classes and the Parkinson's Disease Classification dataset which has 754 output attributes and two output classes. The results indicate that the best-performing architecture for the first dataset is the one that uses one hidden layer, with a PCA cumulative variance of 69.7%, while for the second dataset is the one that uses three hidden layers, with a PCA cumulative variance of 38.9%. Increasing the number of hidden layers does not always improve the performance of neural networks. Therefore, it is essential to determine the number of hidden layers and neurons that are appropriate to achieve good performance in neural networks. The use of PCA and K-Means clustering is expected to provide guidelines in determining neural networks architectures with good performance.

MLP based predictive model for surface ozone concentration over an urban area in the Gangetic West Bengal during pre-monsoon season

The present paper reports a comparative study among two neurocomputing models in the form of Multilayer Perceptron (MLP) models and non-linear regression for the prediction of surface ozone (O3) during pre-monsoon season over Gangetic West Bengal (GWB), India considering NOx, SO2, PM10 and temperature as predictors. Learning the MLPs through gradient descent (GD) with tanhyperbolic and sigmoid nonlinearities, we found that all the models under consideration have almost the same degrees of prediction efficiency for O3 over GWB during pre-monsoon season with the said predictors. However, the MLP model with tanhyperbolic activation function is found to produce a significantly higher correlation and Willmott's index of agreement between actual and predicted O3 than the other models. Finally, MLP with GD learning characterized by tanhyperbolic nonlinearity is identified to have significant efficiency in surface ozone prediction over the region as mentioned above.

Shannon entropy maximization supplemented by neurocomputing to study the consequences of a severe weather phenomenon on some surface parameters

An information theoretic approach based on Shannon entropy is adopted in this study to discern the influence of pre-monsoon thunderstorm on some surface parameters. A few parameters associated with pre-monsoon thunderstorms over a part of east and northeast India are considered. Maximization of Shannon entropy is employed to test the relative contributions of these parameters in creating this weather phenomenon. It follows as a consequence of this information theoretic approach that surface temperature is the most important parameter among those considered. Finally, artificial neural network in the form of multilayer perceptron with backpropagation learning is attempted to develop predictive model for surface temperature.

Nowcasting visibility during wintertime fog over the airport of a metropolis of India: decision tree algorithm and artificial neural network approach

The endeavor of the present research is to nowcast the spatial visibility during fog over the airport of Kolkata (22.6°N; 88.4°E), India, with artificial neural network (ANN) model. The identification of dominant parameters influencing the visibility during wintertime (November–February) fog over the region is made using the decision tree algorithm. The decision tree is constructed by computing the entropy of the parameters collected during the period from 2001 to 2011. The parameters having minimum entropy are selected as the most useful parameters because it has maximum certainty in influencing the visibility. The result reveals that the moderate range of NO2 (67–134 µg/m3) is the most dominant parameter compared with other parameters that influence the visibility during wintertime fog over Kolkata and is selected as the first node of the tree. The decision tree approach led to select five such parameters having minimum entropy for affecting maximum the visibility during fog over Kolkata airport. The selected parameters are NO2, wind speed, relative humidity, CO and temperature. ANN model is developed with the selected parameters as the input in the form of multilayer perceptron with back propagation learning technique for forecasting the 3 hourly visibility during wintertime fog over Kolkata airport. The result reveals that the forecast of visibility of different categories is possible with ANN model. However, the best forecast is obtained for very dense visibility within the 50 m horizontal distance. The result is validated with observation, and the forecast error is estimated.

Multiple regression and Artificial Neural Network for long-term rainfall forecasting using large scale climate modes

• We modelled rainfall using multiple regression and Artificial Neural Networks for Victoria, Australia. • We examined combined lagged months of ENSO and IOD as potential predictors of spring rainfall. • Spring rainfall was predicted three years in advance with the developed models using ENSO–IOD lagged months. • ANN models proved to be more accurate than multiple regression models in terms of long term predictions. In this study, the application of Artificial Neural Networks (ANN) and Multiple regression analysis (MR) to forecast long-term seasonal spring rainfall in Victoria, Australia was investigated using lagged El Nino Southern Oscillation (ENSO) and Indian Ocean Dipole (IOD) as potential predictors. The use of dual (combined lagged ENSO–IOD) input sets for calibrating and validating ANN and MR Models is proposed to investigate the simultaneous effect of past values of these two major climate modes on long-term spring rainfall prediction. The MR models that did not violate the limits of statistical significance and multicollinearity were selected for future spring rainfall forecast. The ANN was developed in the form of multilayer perceptron using Levenberg–Marquardt algorithm. Both MR and ANN modelling were assessed statistically using mean square error (MSE), mean absolute error (MAE), Pearson correlation ( r ) and Willmott index of agreement ( d ). The developed MR and ANN models were tested on out-of-sample test sets; the MR models showed very poor generalisation ability for east Victoria with correlation coefficients of −0.99 to −0.90 compared to ANN with correlation coefficients of 0.42–0.93; ANN models also showed better generalisation ability for central and west Victoria with correlation coefficients of 0.68–0.85 and 0.58–0.97 respectively. The ability of multiple regression models to forecast out-of-sample sets is compatible with ANN for Daylesford in central Victoria and Kaniva in west Victoria ( r = 0.92 and 0.67 respectively). The errors of the testing sets for ANN models are generally lower compared to multiple regression models. The statistical analysis suggest the potential of ANN over MR models for rainfall forecasting using large scale climate modes.

A neurocomputing approach to the forecasting of monthly maximum temperature over Kolkata, India using total ozone concentration as predictor

The association between the monthly total ozone concentration and monthly maximum temperature over Kolkata (22.56° N, 88.30° E), India, has been explored in this paper. For this, the predictability of monthly maximum temperature based on the total ozone as predictor is investigated using Artificial Neural Network. The presence of persistence and similar cyclic patterns are revealed through autocorrelation and cross-correlation coefficients. Common cycles of length 12 and 6 have been identified through periodogram. Hence, a predictive model has been generated by Artificial Neural Network in the form of Multi Layer Perceptron (MLP) using scaled conjugate gradient learning with sigmoid non-linearity. After training and testing the network, an MLP with total ozone of month n as predictor and maximum temperature of month ( n + 1) as the target output is found as the best model. Performance of the model has been judged statistically. Finally, the MLP model has been compared with linear and non-linear regressions and the efficiency of MLP has been established over the regression models.

Identification of the best architecture of a multilayer perceptron in modeling daily total ozone concentration over Kolkata, India

Autoregressive neural network (AR-NN) models of various orders have been generated in this work for the daily total ozone (TO) time series over Kolkata (22.56°N, 88.5°E). Artificial neural network in the form of multilayer perceptron (MLP) is implemented in order to generate the AR-NN models of orders varying from 1 to 13. An extensive variable selection method through multiple linear regression (MLR) is implemented while developing the AR-NNs. The MLPs are characterized by sigmoid non-linearity. The optimum size of the hidden layer is identified in each model and prediction are produced by validating it over the test cases using the coefficient of determination (R 2) and Willmott’s index (WI). It is observed that AR-NN model of order 7 having 6 nodes in the hidden layer has maximum prediction capacity. It is further observed that any increase in the orders of AR-NN leads to less accurate prediction.

التعامل مع التعقيدات الزلزالية باستخدام التقنيات الحاسوبية العصبية وتحديد قدرها الزلزالي ببعض المتوقعات منخفضة المقارنة

Earthquake, being one of the most hazardous geophysical events, has fascinated the scientists all over the globe to investigate an efficient predictive methodology. In India, earthquake is not a rare event. Purpose of the present paper is to view the earthquake as a complex system and then to estimate the magnitude of earthquake over Indian subcontinent using artificial neural network with backpropagation learning. The day, month, and year of occurrence of the earthquake, latitude of the place, and the longitude of the place are considered as the predictor variables. After considering various sizes of the hidden layer, it has been found that artificial neural network in the form of multilayer perceptron with two hidden nodes and sigmoid nonlinearity can be used to anticipate the magnitude of earthquake over Indian subcontinent. The paper concludes with an outline for future work.

Artificial neural network with backpropagation learning to predict mean monthly total ozone in Arosa, Switzerland

The present study deals with the mean monthly total ozone time series over Arosa, Switzerland. The study period is 1932–1971. First of all, the total ozone time series has been identified as a complex system and then Artificial Neural Network models in the form of Multilayer Perceptron with back propagation learning have been developed. The models are Single‐hidden‐layer and Two‐hidden‐layer Perceptrons with sigmoid activation function. After sequential learning with a learning rate of 0.9 the peak total ozone period (February–May) concentrations of mean monthly total ozone have been predicted by the two neural net models. After training and validation, both of the models are found to be skillful. But the Two‐hidden‐layer Perceptron is found to be more adroit in predicting the mean monthly total ozone concentrations over the aforesaid period.

A hybrid approach of NN and HMM for facial emotion classification

Neural networks (NNs) are often combined with Hidden Markov Models (HMMs) in speech recognition for achieving superior performance. In this paper, this hybrid approach is employed in facial emotion classification. Gabor wavelets are employed to extract features from difference images obtained by subtracting the first frame showing a frontal face from the current frame. The NN, which takes the form of Multilayer perceptron (MLP), is used to classify the feature vector into different states of a HMM of a certain emotion sequence, i.e., neutral, intermediate and peak. In addition to using 1–0 as targets for the NN, a heuristic strategy of assigning variable targets 1– x–0 has also been applied. After training, we interpret the output values of the NN as the posterior of the HMM state and directly apply the Viterbi algorithm to these values to estimate the best state path. The experiments show that with variable targets for the NN, the HMM gives better results than that with 1–0 targets. The best HMM results are obtained for x=0.8 in 1– x–0.