Search

Your search keyword '"Jia, Meiying"' showing total 24 results
Start Over Author "Jia, Meiying" Search Limiters Available in Library Collection
24 results on '"Jia, Meiying"'

23. Mutual Information-Based Modified Randomized Weights Neural Networks

Author

Jian Tang, Jia Meiying, Zhuo Liu, and Zhiwei Wu

Subjects
Probability theory, Artificial neural network, Generalization, business.industry, Activation function, Pattern recognition, Feature selection, Mutual information, Artificial intelligence, Echo state network, business, Randomized algorithm, Mathematics
Abstract

Randomized weights neural networks have fast learning speed and good generalization performance with one single hidden layer structure. Input weighs of the hidden layer are produced randomly. By employing certain activation function, outputs of the hidden layer are calculated with some randomization. Output weights are computed using pseudo inverse. Mutual information can be used to measure mutual dependence of two variables quantitatively based on the probability theory. In this paper, these hidden layer’s outputs that relate to prediction variable closely are selected with the simple mutual information based feature selection method. These hidden nodes with high mutual information values are maintained as a new hidden layer. Thus, the size of the hidden layer is reduced. The new hidden layer’s output weights are learned with the pseudo inverse method. The proposed method is compared with the original randomized algorithms using concrete compressive strength benchmark dataset.

Published
2015

24. Analysis Method of Multi-scale Frequency Spectral Latent Feature based on the Mutual Information

Author

Liu Zhuo, Zhou Xiaojie, Jia Meiying, and Jian Tang

Subjects
Vibration, Engineering, Scale (ratio), business.industry, Kernel (statistics), Feature extraction, Partial least squares regression, Principal component analysis, Process (computing), Pattern recognition, Artificial intelligence, Mutual information, business
Abstract

Heavy rotating mechanical devices, such as ball mills in the mineral grinding process, produce strong mechanical vibration and acoustic signals. These signals consist of multiple components with different time scales and physical interpretations. Frequency spectra of these multi-scale sub-signals contain interesting information related to some difficulty-to-measure process parameters. Latent feature extraction methods based on principal component analysis and partial least squares with their kernel versions are used widely for high dimensional spectra data. How to estimate their effect and select important sub-signals are important issues. Thus, aim to these problems, mutual information is used in this paper. Analysis results based on the shell vibration frequency spectra of a laboratory-scale wet ball mill shows that the proposed method is helpful and effective.

Published
2015

Catalog

Books, media, physical & digital resources
See catalog results