Analysis of the Potential in Convolutional Deep Independent Component Analysis

DSpace Repository

Show simple item record

dc.contributor.advisor Hong, Pengyu en_US
dc.contributor.author Jin, Zexi
dc.date.accessioned 2019-05-13T13:17:09Z
dc.date.available 2019-05-13T13:17:09Z
dc.date.issued 2018 en_US
dc.identifier.uri https://hdl.handle.net/10192/36633
dc.description.abstract In this paper, the prospects and potentials of Convolutional Deep Independent Component Analysis are analyzed. Convolutional Deep Independent Component Analysis, abbreviate as CDICA, is an technique of incorporating Independent Component Analysis(ICA) with Fully Connected Neural Networks(FCNN). CDICA aims to simulate the convolutions similar the Convolutional Neural Network(CNN) with less time, less hyperparameters, and much smaller training samples. The simulation of convolutions are done by ICA algorithms and the feature merging step. The validity of CDICA are tested in this proposal. en_US
dc.language English
dc.language.iso en en_US
dc.relation.ispartofseries Brandeis University Theses and Dissertations
dc.rights Copyright by Zexi Jin, 2018 en_US
dc.title Analysis of the Potential in Convolutional Deep Independent Component Analysis en_US
dc.type Thesis en_US
dc.contributor.department Department of Computer Science
dc.degree.name BS
dc.degree.level Bachelors
dc.degree.discipline Computer Science


Files in this item

This item appears in the following Collection(s)

Show simple item record

Search BIR


Browse

My Account