Font Size: a A A

Neural Network Ensemble Research And Its Application In The Coal Mine Safety

Posted on:2013-04-03Degree:MasterType:Thesis
Country:ChinaCandidate:B B LiFull Text:PDF
GTID:2248330371969681Subject:Management Science and Engineering
Abstract/Summary:PDF Full Text Request
Since the concept of neural network ensemble was proposed, a large number ofresearchers have been entering this field, which makes the neural network ensemblebecome hot and popular in the international machine learning and neural computationacademia area. As a result, how to design more effective neural network ensemblemethod, which can improve the generalization ability of neural network ensemble, isof great theoretical and practical significance.In this article, we will propose new algorithms respectively in the sample-selectionstage, the individual network generation and selection stage under this premise. Therewill be a new sample selection algorithm based on graph clustering in the sample-selection stage. In the individual network generation and selection stage, I willpropose a network of individual selection algorithm selection algorithm, which isbased on artificial bee colony.Chapterâ…¢focuses on the graph-clustering algorithm. I will apply thegraph-clustering algorithm to training sample selection, and propose a neural networkensemble algorithm based on graph-clustering. First, cluster sample based ongraph-clustering algorithm. Second, select sample from the sub-clustering. And then,train individual neural networks from the sample. At last, integrate these Individualneural networks.In the fourth part, there will be a selective neural network ensemble algorithmbased on ABC algorithm (swarm algorithm), which can remove redundant individualand build integrated network of high accuracy during great different integration ofindividuals. The algorithm can train a large number of integrated individual repeatedly.And then, use the swarm algorithm to make the prediction error of networkintegration as the optimization objective function. And integrate selectivesub-individuals finally. In the end, this article focuses on the application of the technology of neuralnetwork ensemble in the coal mine safety hazards field, especially in prediction of gasemission in coal mines.
Keywords/Search Tags:Neural network ensemble, Graph-clustering, Bees Algorithm, Coal mine safe
PDF Full Text Request
Related items