Using Stochastic Encoders to Discover Structure in Data

dc.creatorLuttrell, Stephen
dc.date2004-08-21
dc.date.accessioned2026-07-07T03:21:42Z
dc.date.available2026-07-07T03:21:42Z
dc.descriptionIn this paper a stochastic generalisation of the standard Linde-Buzo-Gray (LBG) approach to vector quantiser (VQ) design is presented, in which the encoder is implemented as the sampling of a vector of code indices from a probability distribution derived from the input vector, and the decoder is implemented as a superposition of reconstruction vectors. This stochastic VQ (SVQ) is optimised using a minimum mean Euclidean reconstruction distortion criterion, as in the LBG case. Numerical simulations are used to demonstrate how this leads to self-organisation of the SVQ, where different stochastically sampled code indices become associated with different input subspaces.
dc.description18 pages, 9 figures. Full version of a short paper that was published in the Digest of the 5th IMA International Conference on Mathematics in Signal Processing, 18-20 December 2000, Warwick University, UK
dc.identifierhttps://arxiv.org/abs/cs/0408049
dc.identifierhttp://arxiv.org/abs/cs/0408049
dc.identifier.urihttp://salesiana.dossiersoluciones.com/handle/123456789/32304
dc.subjectNeural and Evolutionary Computing
dc.subjectComputer Vision and Pattern Recognition
dc.subjectI.2.6; I.5.1
dc.titleUsing Stochastic Encoders to Discover Structure in Data
dc.typetext

Files

Collections