Publication Details
Self-Organizing Sparse Distributed Memory as a Predictive Memory
Neural Net, Self-Organizing Map, Soft Competitive Learning Rule, Sparse Distributed Memory, Prediction
The paper discuses an extension of Kanerva's Sparse Distributed Memory (SDM) and introduces possible application in prediction. A self-organizing SDM equivalent to a three-layered neural network is used to learn the desired transfer function mapping a history of velocity vectors to predicted velocity vector. The net was tested in simple mouse-tracking experiment, but can be used in more handy problems, for example in motion capturing systems.
The paper discuses an extension of Kanerva's Sparse Distributed Memory (SDM) and introduces possible application in prediction. A self-organizing SDM equivalent to a three-layered neural network is used to learn the desired transfer function mapping a history of velocity vectors to predicted velocity vector. The net was tested in simple mouse-tracking experiment, but can be used in more handy problems, for example in motion capturing systems.
@INPROCEEDINGS{FITPUB5787, author = "Franti\v{s}ek Greben\'{i}\v{c}ek", title = "Self-Organizing Sparse Distributed Memory as a Predictive Memory", pages = "17--22", booktitle = "Nostradamus '99", year = 1999, location = "Zl\'{i}n, CZ", ISBN = "80-214-1424-3", language = "english", url = "https://www.fit.vut.cz/research/publication/5787" }