Calibration of Constraint Promotion Does Not Help with Learning Variation in Stochastic Optimality Theory

Bibliographic Data

ID10719344
AuthorsGiorgio Magri (0000-0003-0350-9235, SFL UMR 7023 (UPL, CNRS, University of Paris 8), 61 rue Pouchet, 75017 Paris, France, corresponding author), Benjamin Storme (0000-0002-2560-2265, SFL UMR 7023 (UPL, CNRS, University of Paris 8), 61 rue Pouchet, 75017 Paris, France, corresponding author)
Year2020
Volume51
Issue1
Pages97-123
Publication date2020-01-01
Open AccessNo
TypeARTICLE
VenueLinguistic Inquiry (JOURNAL)
Journal identifiersISSN: 0024-3892 • E-ISSN: 1530-9150
PublisherMIT Press - Journals (PUBLISHER)
DOI10.1162/ling_a_00328
OpenAlexW2903500617
LanguageEN
References cited9

The Calibrated Error-Driven Ranking Algorithm (CEDRA; Magri 2012 ) is shown to fail on two test cases of phonologically conditioned variation from Boersma and Hayes 2001 . The failure of the CEDRA raises a serious unsolved challenge for learnability research in stochastic Optimality Theory, because the CEDRA itself was proposed to repair a learnability problem ( Pater 2008 ) encountered by the original Gradual Learning Algorithm. This result is supported by both simulation results and a detailed analysis whereby a few constraints and a few candidates at a time are recursively “peeled off” until we are left with a “core” small enough that the behavior of the learner is easy to interpret

Algorithm · Calibration · Constraint (computer-aided design) · Core (optical fiber) · Learnability · Linguistics · Machine learning · Mathematical economics · Mathematical optimization · Natural language processing · Optimality theory · Phonology · Promotion (chess) · Ranking (information retrieval) · Statistics · Variation (astronomy) · Advanced Bandit Algorithms Research · Artificial Intelligence · Computer Science · Machine Learning and Algorithms · Mathematics · Neural Networks and Applications · Philosophy

Tools

Open DOISci-Hub