How to keep the HG weights non-negative: the truncated Perceptron reweighing rule

The literature on error-driven learning in Harmonic Grammar (HG) has adopted the Perceptron reweighing rule. Yet, this rule is not suited to HG, as it fails at ensuring non-negative weights. A variant is thus considered which truncates the updates at zero, keeping the weights non-negative. Converge...

Full description

Bibliographic Details
Main Author:	Giorgio Magri
Format:	Article
Language:	English
Published:	Polish Academy of Sciences 2015-12-01
Series:	Journal of Language Modelling
Subjects:	Harmonic Grammar error-driven learning Perceptron convergence
Online Access:	https://jlm.ipipan.waw.pl/index.php/JLM/article/view/115

id	doaj-0ce5d491b88e440c9d4f02ceca0695da
record_format	Article
spelling	doaj-0ce5d491b88e440c9d4f02ceca0695da2021-02-25T14:51:00ZengPolish Academy of SciencesJournal of Language Modelling2299-856X2299-84702015-12-013210.15398/jlm.v3i2.11540How to keep the HG weights non-negative: the truncated Perceptron reweighing ruleGiorgio Magri0UMR 7023 SFL (CNRS, University of Paris 8) The literature on error-driven learning in Harmonic Grammar (HG) has adopted the Perceptron reweighing rule. Yet, this rule is not suited to HG, as it fails at ensuring non-negative weights. A variant is thus considered which truncates the updates at zero, keeping the weights non-negative. Convergence guarantees and error bounds for the original Perceptron are shown to extend to its truncated variant. https://jlm.ipipan.waw.pl/index.php/JLM/article/view/115Harmonic Grammarerror-driven learningPerceptronconvergence
collection	DOAJ
language	English
format	Article
sources	DOAJ
author	Giorgio Magri
spellingShingle	Giorgio Magri How to keep the HG weights non-negative: the truncated Perceptron reweighing rule Journal of Language Modelling Harmonic Grammar error-driven learning Perceptron convergence
author_facet	Giorgio Magri
author_sort	Giorgio Magri
title	How to keep the HG weights non-negative: the truncated Perceptron reweighing rule
title_short	How to keep the HG weights non-negative: the truncated Perceptron reweighing rule
title_full	How to keep the HG weights non-negative: the truncated Perceptron reweighing rule
title_fullStr	How to keep the HG weights non-negative: the truncated Perceptron reweighing rule
title_full_unstemmed	How to keep the HG weights non-negative: the truncated Perceptron reweighing rule
title_sort	how to keep the hg weights non-negative: the truncated perceptron reweighing rule
publisher	Polish Academy of Sciences
series	Journal of Language Modelling
issn	2299-856X 2299-8470
publishDate	2015-12-01
description	The literature on error-driven learning in Harmonic Grammar (HG) has adopted the Perceptron reweighing rule. Yet, this rule is not suited to HG, as it fails at ensuring non-negative weights. A variant is thus considered which truncates the updates at zero, keeping the weights non-negative. Convergence guarantees and error bounds for the original Perceptron are shown to extend to its truncated variant.
topic	Harmonic Grammar error-driven learning Perceptron convergence
url	https://jlm.ipipan.waw.pl/index.php/JLM/article/view/115
work_keys_str_mv	AT giorgiomagri howtokeepthehgweightsnonnegativethetruncatedperceptronreweighingrule
_version_	1724251373656080384

How to keep the HG weights non-negative: the truncated Perceptron reweighing rule

Similar Items