Author(s): G. Jananii
Journal: Bonfring International Journal of Man Machine Interface
ISSN 2250-1061
Volume: 02;
Issue: 04;
Start page: 01;
Date: 2012;
VIEW PDF
DOWNLOAD PDF 
Original page
Keywords: Neural Network | Large Scale Dataset | Incremental Learning
ABSTRACT
Investigation on large data sets is extremely important in data mining.Large amount of data generally requires a specific learning method or of any optimization method. Particularly some standard methods are used for example Artificial Neural Network, back propagation neural network and other neural networksnecessitate very long learning time.The existing technique that does not performed well on the large data sets. So in this paper presents a new approach called multi layered feed forward neural network which can work efficiently with the neural networks on large data sets.Data is separated into several segments, and learned by anidentical network structure whereas all weights from the set of networks are integrated. The results from the experiments show that the proposed method can protect the accuracy while the training time is significantly reduced.
Journal: Bonfring International Journal of Man Machine Interface
ISSN 2250-1061
Volume: 02;
Issue: 04;
Start page: 01;
Date: 2012;
VIEW PDF
DOWNLOAD PDF Original pageKeywords: Neural Network | Large Scale Dataset | Incremental Learning
ABSTRACT
Investigation on large data sets is extremely important in data mining.Large amount of data generally requires a specific learning method or of any optimization method. Particularly some standard methods are used for example Artificial Neural Network, back propagation neural network and other neural networksnecessitate very long learning time.The existing technique that does not performed well on the large data sets. So in this paper presents a new approach called multi layered feed forward neural network which can work efficiently with the neural networks on large data sets.Data is separated into several segments, and learned by anidentical network structure whereas all weights from the set of networks are integrated. The results from the experiments show that the proposed method can protect the accuracy while the training time is significantly reduced.
