Two Strategies to Avoid Overfitting in Feedforward Networks

Schittenkopf, Christian; Deco, Gustavo; Brauer, Wilfried

doi:10.1016/s0893-6080(96)00086-x

Cited by 115 publications

(45 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Multilayer perceptron (MLP) neural networks are universal function approximators (Lek & Guegan, 1999). With enough twisting of the parameters, one can use them to implement any classification rule (Schittenkopf, Deco, & Brauer, 1997). Consequently, the networks could in theory also be designed to explain the version of the dataset in which the GPA scores would be randomly assigned to the students.…”

Section: Fitting To the Test Datamentioning

confidence: 99%

Using New Models to Analyze Complex Regularities of the World

Nokelainen

2014

FLR

View full text Add to dashboard Cite

show abstract

Section: Fitting To the Test Datamentioning

confidence: 99%

Using New Models to Analyze Complex Regularities of the World

Nokelainen

2014

FLR

View full text Add to dashboard Cite

show abstract

“…In general, this type of modeling works well if the training data presented to the ANFIS is fully representative of the features of the data the trained system is intended to model. Checking and testing data sets allow checking the generalization capability of the resulting fuzzy inference system and avoiding overfitting [51].…”

Section: Anfis Architecturementioning

confidence: 99%

Artificial intelligence tools and inverse methods for estimating the thermal diffusivity of building materials

Grieu

Faugeroux

Traoré

et al. 2011

Energy and Buildings

View full text Add to dashboard Cite

the actual European energy context highlights the building sector as one of the largest sectors of energy consumption. Consequently, the "Energy Performance of Buildings Directive", adopted in 2002 and focusing on energy use in buildings, requires all the EU members to enhance their building regulations and to introduce energy certification schemes, with the aim of both reducing energy consumption and improving energy efficiency. That is why carrying out an energy performance diagnosis is mandatory, notably when buying or selling properties. Indeed, invisible defaults, like, for example, non-emerging cracks or delaminations, could have a detrimental effect on insulating qualities. Esimaing in-situ thermo-physical properties allowing locating these defaults, the present work focuses on proposing new and efficient approaches based on the use of both artificial intelligence tools (artificial neural networks and neuro-fuzzy systems) and inverse methods for characterizing building materials i.e. for estimating their thermal diffusivity using thermograms obtained thanks to a non-destructive photothermal method.

show abstract

“…One of this research objectives is to prevent overfitting (Schittenkopf et al, 1997). The prediction procedure thus utilizes a PCA and a K-Means clustering in order to optimize the network training.…”

Section: Neural Network Structurementioning

confidence: 99%

“…So the network will be able to focus on the important features in the data rather than fitting the noise, an inherent component of any environmental field data set (Schittenkopf et al, 1997;El-Din and Smith, 2002).…”

Section: Article In Pressmentioning

confidence: 99%

Prediction of parameters characterizing the state of a pollution removal biologic process

Grieu¹,

Traoré²,

Polit³

et al. 2005

Engineering Applications of Artificial Intelligence

View full text Add to dashboard Cite

International audienceThis work is devoted to the prediction, based on neural networks, of physicochemical parameters impossible to measure on-line. These parameters-the Chemical Oxygen Demand (COD) and the ammonia NH 4-characterize the organic matter and nitrogen removal biological process carried out at the Saint Cyprien WWTP (France). Their knowledge make it possible to estimate the process quality and efficiency. First, the data are treated by K-Means clustering then by principal components analysis in order to optimize the Multi-Level Perceptron (MLP) learning phase. K-Means clustering makes it possible to highlight different operations within the Saint Cyprien treatment plant. The Principal Components Analysis (PCA) is used to eliminate redundancies and synthesizes the information expressed by a data set. With respect to the neural network used, these techniques facilitate the pollution removal process understanding and the identification of existing relations between the predictive variables and the variables to be predicted

show abstract

Two Strategies to Avoid Overfitting in Feedforward Networks

Cited by 115 publications

References 12 publications

Using New Models to Analyze Complex Regularities of the World

Using New Models to Analyze Complex Regularities of the World

Artificial intelligence tools and inverse methods for estimating the thermal diffusivity of building materials

Prediction of parameters characterizing the state of a pollution removal biologic process

Contact Info

Product

Resources

About