Please use this identifier to cite or link to this item:
Title: Dynamics analysis and applications of neural networks
Keywords: feed-forward neural networks, recurrent networks, dynamics analysis, quadratic optimization, combinatorial optimization, linear threshold networks
Issue Date: 17-Feb-2005
Citation: TANG HUAJIN (2005-02-17). Dynamics analysis and applications of neural networks. ScholarBank@NUS Repository.
Abstract: This thesis comprises several important issues in learning theory, dynamics analysis and applications of feed-forward neural networks, recurrent networks with saturating transfer functions and with nonsaturating transfer functions. To overcome sensitivity and slow learning of conventional back-propagation algorithm, a new training algorithm for multilayer feed-forward neural networks is put forward. A discrete-time recurrent network is proposed as well as the conditions on the global exponential stability. To improve the solution quality of combinatorial optimization problems, as another important application of Hopfield network, a new parameter setting rule is set up. The competitive model incorporating winner-take-all mechanism is presented which is capable of eliminating tedious process of parameter settings and increasing computation efficiency significantly. One important focus of this thesis is on the linear threshold (LT) networks, of which the transfer functions are nonsaturating. Various dynamical properties are clarified and new theoretical results are acquired which facilitate their applications, such as the analog associative memory.
Appears in Collections:Ph.D Theses (Open)

Show full item record
Files in This Item:
File Description SizeFormatAccess SettingsVersion 
HT016646Y_PhD.pdf1.92 MBAdobe PDF



Page view(s)

checked on Oct 7, 2018


checked on Oct 7, 2018

Google ScholarTM


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.