A comparison of methods to avoid overfitting in neural networks training in the case of catchment runoff modelling

Piotrowski, A.; Napiorkowski, Jarosław J.

doi:10.1016/j.jhydrol.2012.10.019

Cited by 221 publications

(92 citation statements)

References 82 publications

(108 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…They are connected by the so-called synapses which are related with appropriate weighting factors. The most commonly used network model is the three-layer ANN model, in which we distinguish the input layer, hidden layer and output layer [38][39][40][41][42][43][44][45][46].…”

Section: Artificial Neural Network (Ann)mentioning

confidence: 99%

“…To define the final model, time series regression analysis was performed using the multilayer perceptron (MLP) ANN model with the maximum number of hidden layers defined as 10, and linear, logistic, tanh, exponential, and sinusoidal functions were used as activation functions for hidden and output neurons [38][39][40][41][42][43][44][45][46].…”

Section: Artificial Neural Network (Ann)mentioning

confidence: 99%

“…In this study, the results obtained using ANN were compared and evaluated [44][45][46][47][48][49][50][51]. It is important to define average forecast error by (MSE, RMSE), model fit (R 2 ), and prognostic error distribution when creating forecast models.…”

Section: Model Performance Indicatorsmentioning

confidence: 99%

“…It is important to define average forecast error by (MSE, RMSE), model fit (R 2 ), and prognostic error distribution when creating forecast models. The robustness of the model was tested by determining the mean absolute error of the prediction (MAPE) -defining the percent precision of the model according to the formula [44][45][46][47][48][49][50][51] …”

Section: Model Performance Indicatorsmentioning

confidence: 99%

See 3 more Smart Citations

Prediction of the Seasonal Changes of the Chloride Concentrations in Urban Water Reservoir

Miller

Poleszczuk

2017

Ecological Chemistry and Engineering S

View full text Add to dashboard Cite

This study investigated the possibility of using artificial neural networks to predict changes in the concentration of chloride ions in the urban ponds on the example of the inflow and outflow zones of water to and from the ponds Syrenie Stawy in Szczecin (NW-Poland). The possibility of using selected water quality indices (selected based on correlation matrix of water quality indices with Cl -), in particular: COD-Cr, BOD5, DO, water saturation by O2 and NO2 -and their influence on the chloride concentration forecast was tested.

show abstract

Section: Artificial Neural Network (Ann)mentioning

confidence: 99%

Section: Artificial Neural Network (Ann)mentioning

confidence: 99%

Section: Model Performance Indicatorsmentioning

confidence: 99%

Section: Model Performance Indicatorsmentioning

confidence: 99%

See 2 more Smart Citations

Prediction of the Seasonal Changes of the Chloride Concentrations in Urban Water Reservoir

Miller

Poleszczuk

2017

Ecological Chemistry and Engineering S

View full text Add to dashboard Cite

show abstract

“…Overfitting condition causes poor performance, and therefore modelling needs to be done again to get a good fit condition [4]. Underfitting refers to a condition when an algorithm is not able to model itself from training set, or cannot generalize itself to a new data.…”

Section: Polynomial Regression Polynomial Regression Is a Variantmentioning

confidence: 99%

Review of Regression Analysis Models

Gupta¹,

Sharma²,

Goel³

2017

IJERT

View full text Add to dashboard Cite

Abstract-In statistics and data analysis, we often need to establish a relationship between the various parameters in a data set. This relationship is important for prediction and analysis. Regression Analysis is such a technique. This work mainly focuses on the different Regression Analysis models used nowadays and how they are used in context of different data sets. Picking the right model for analysis is often the most difficult task and therefore, these models are looked upon closely in this research. While a Linear Regression Analysis model is used to fit linear data, a Polynomial Regression Analysis model focuses on a data set representing polynomial relationship between data parameters. Logistic Regression model is used in a scenario where we need a binary type of prediction. When the data set becomes complex, these models may suffer from issues like Underfitting and Overfitting. Ridge and Lasso Regression are considered the best models to deal with this type of situation. Ridge regression is used when data suffers from multicollinearity, that is independent variables are highly correlated. Lasso regression differs from ridge regression in a way that it uses absolute values in the penalty function, instead of squares. Using these models in the right way and with right data set, Data Analysis and Prediction can produce the most accurate results.

show abstract

Artificial Intelligence Model for Forecasting of Membrane Fouling in Wastewater Treatment by Membrane Technology

Schmitt

2020

Modeling in Membranes and Membrane‐Based Processes

View full text Add to dashboard Cite

A comparison of methods to avoid overfitting in neural networks training in the case of catchment runoff modelling

Cited by 221 publications

References 82 publications

Prediction of the Seasonal Changes of the Chloride Concentrations in Urban Water Reservoir

Prediction of the Seasonal Changes of the Chloride Concentrations in Urban Water Reservoir

Review of Regression Analysis Models

Artificial Intelligence Model for Forecasting of Membrane Fouling in Wastewater Treatment by Membrane Technology

Contact Info

Product

Resources

About