dc.contributor.authorLin, Xiang
dc.date.accessioned2018-09-24T06:09:22Z
dc.date.available2018-09-24T06:09:22Z
dc.date.issued2018
dc.identifier.urihttp://hdl.handle.net/10356/76036
dc.description.abstractWith the booming development of machine learning and deep learning, Artificial Intelligence has achieved noticeable progress nowadays. More and more complex problems are solved by utilizing machine learning technique which requires deeper architecture of the artificial neural network to deal with large-scale data information and complicated scenario. The Random Vector Functional Link (RVFL) neural network is a universal approximator that has been applied to many areas to solve practical problems. However, a multi-layer architecture of RVFL has not yet been explored before. The proposedMulti-layer RVFL consists of two parts, the classifier that serves as the same function as single hidden layer RVFL, and the feature extraction part that extracts more meaningful information of input data for further classification. The feature extraction part consists of several RVFL based auto-encoders to exploit the random mapping capability of RVFL. The RVFL based auto-encoder does not need an iterative computation due to its closed-form solution which is faster to compute than gradient based back propagation method. Extensive experiments are conducted to demonstrate the great performance of Multi-layer RVFL compared to single hidden layer RVFL.en_US
dc.format.extent55 p.en_US
dc.language.isoenen_US
dc.subjectDRNTU::Engineering::Electrical and electronic engineeringen_US
dc.titleMultilayer random vector functional link neural networksen_US
dc.typeThesis
dc.contributor.supervisorPonnuthurai N. Suganthanen_US
dc.contributor.schoolSchool of Electrical and Electronic Engineeringen_US
dc.description.degreeMaster of Science (Computer Control and Automation)en_US


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record