{"id":"https://openalex.org/W4412945124","doi":"https://doi.org/10.18653/v1/2025.acl-long.1180","title":"GRaMPa: Subword Regularisation by Skewing Uniform Segmentation Distributions with an Efficient Path-counting Markov Model","display_name":"GRaMPa: Subword Regularisation by Skewing Uniform Segmentation Distributions with an Efficient Path-counting Markov Model","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412945124","doi":"https://doi.org/10.18653/v1/2025.acl-long.1180"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.acl-long.1180","is_oa":false,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.1180","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://lirias.kuleuven.be/handle/20.500.12942/779656","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5114337210","display_name":"Thomas Bauwens","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Thomas Bauwens","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5119204451","display_name":"David Kacz\u00e9r","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"David Kacz\u00e9r","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5080895973","display_name":"Miryam de Lhoneux","orcid":"https://orcid.org/0000-0001-8844-2126"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Miryam de Lhoneux","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.09671872,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"24228","last_page":"24257"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.9991000294685364,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.9991000294685364,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10201","display_name":"Speech Recognition and Synthesis","score":0.9939000010490417,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11550","display_name":"Text and Document Classification Technologies","score":0.9750999808311462,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.6714550256729126},{"id":"https://openalex.org/keywords/path","display_name":"Path (computing)","score":0.6440749168395996},{"id":"https://openalex.org/keywords/markov-chain","display_name":"Markov chain","score":0.6413177251815796},{"id":"https://openalex.org/keywords/markov-process","display_name":"Markov process","score":0.5496382713317871},{"id":"https://openalex.org/keywords/segmentation","display_name":"Segmentation","score":0.5343097448348999},{"id":"https://openalex.org/keywords/hidden-markov-model","display_name":"Hidden Markov model","score":0.5229811668395996},{"id":"https://openalex.org/keywords/markov-model","display_name":"Markov model","score":0.5050056576728821},{"id":"https://openalex.org/keywords/image-segmentation","display_name":"Image segmentation","score":0.4848552942276001},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.4190168082714081},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.3816995322704315},{"id":"https://openalex.org/keywords/pattern-recognition","display_name":"Pattern recognition (psychology)","score":0.3650604784488678},{"id":"https://openalex.org/keywords/speech-recognition","display_name":"Speech recognition","score":0.3323047161102295},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.2728634774684906},{"id":"https://openalex.org/keywords/statistics","display_name":"Statistics","score":0.20959243178367615},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.16437190771102905},{"id":"https://openalex.org/keywords/computer-network","display_name":"Computer network","score":0.05899810791015625}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6714550256729126},{"id":"https://openalex.org/C2777735758","wikidata":"https://www.wikidata.org/wiki/Q817765","display_name":"Path (computing)","level":2,"score":0.6440749168395996},{"id":"https://openalex.org/C98763669","wikidata":"https://www.wikidata.org/wiki/Q176645","display_name":"Markov chain","level":2,"score":0.6413177251815796},{"id":"https://openalex.org/C159886148","wikidata":"https://www.wikidata.org/wiki/Q176645","display_name":"Markov process","level":2,"score":0.5496382713317871},{"id":"https://openalex.org/C89600930","wikidata":"https://www.wikidata.org/wiki/Q1423946","display_name":"Segmentation","level":2,"score":0.5343097448348999},{"id":"https://openalex.org/C23224414","wikidata":"https://www.wikidata.org/wiki/Q176769","display_name":"Hidden Markov model","level":2,"score":0.5229811668395996},{"id":"https://openalex.org/C163836022","wikidata":"https://www.wikidata.org/wiki/Q6771326","display_name":"Markov model","level":3,"score":0.5050056576728821},{"id":"https://openalex.org/C124504099","wikidata":"https://www.wikidata.org/wiki/Q56933","display_name":"Image segmentation","level":3,"score":0.4848552942276001},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4190168082714081},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3816995322704315},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.3650604784488678},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.3323047161102295},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.2728634774684906},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.20959243178367615},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.16437190771102905},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.05899810791015625}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.18653/v1/2025.acl-long.1180","is_oa":false,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.1180","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},{"id":"pmh:oai:lirias2repo.kuleuven.be:20.500.12942/779656","is_oa":true,"landing_page_url":"https://lirias.kuleuven.be/handle/20.500.12942/779656","pdf_url":null,"source":{"id":"https://openalex.org/S7407055369","display_name":"Lirias","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"acceptedVersion","is_accepted":true,"is_published":false,"raw_source_name":"63rd Annual Meeting of the Association for Computational Linguistics, Vienna, Austria","raw_type":"info:eu-repo/semantics/conferenceObject"}],"best_oa_location":{"id":"pmh:oai:lirias2repo.kuleuven.be:20.500.12942/779656","is_oa":true,"landing_page_url":"https://lirias.kuleuven.be/handle/20.500.12942/779656","pdf_url":null,"source":{"id":"https://openalex.org/S7407055369","display_name":"Lirias","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"acceptedVersion","is_accepted":true,"is_published":false,"raw_source_name":"63rd Annual Meeting of the Association for Computational Linguistics, Vienna, Austria","raw_type":"info:eu-repo/semantics/conferenceObject"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W1510894296","https://openalex.org/W2134386692","https://openalex.org/W2379651310","https://openalex.org/W2113019827","https://openalex.org/W1541249122","https://openalex.org/W2084326697","https://openalex.org/W2194396582","https://openalex.org/W2027903142","https://openalex.org/W2082284720","https://openalex.org/W2116722627"],"abstract_inverted_index":{"Stochastically":[0],"sampling":[1,69,134],"word":[2],"segmentations":[3,53],"from":[4,104],"a":[5,13,48,178],"subword":[6,10,42],"tokeniser,":[7],"also":[8],"called":[9],"regularisation,":[11],"is":[12,150],"known":[14],"way":[15],"to":[16,22,54,61,142],"increase":[17],"robustness":[18],"of":[19,50,70,91,100],"language":[20,168],"models":[21,169],"out-of-distribution":[23],"inputs,":[24],"such":[25,184],"as":[26,185],"text":[27],"containing":[28],"spelling":[29,190],"errors.":[30],"Recent":[31],"work":[32],"has":[33,58],"observed":[34],"that":[35,38,82,167],"usual":[36],"augmentations":[37],"make":[39],"popular":[40],"deterministic":[41],"tokenisers":[43,176],"stochastic":[44],"still":[45,107],"cause":[46],"only":[47],"handful":[49],"all":[51],"possible":[52],"be":[55],"sampled.":[56],"It":[57],"been":[59],"proposed":[60],"uniformly":[62,83],"sample":[63],"across":[64],"these":[65],"instead,":[66],"through":[67,131],"rejection":[68,162],"paths":[71,130],"in":[72,86,177],"an":[73,116],"unweighted":[74],"segmentation":[75,85,123],"graph.":[76],"In":[77],"this":[78],"paper,":[79],"we":[80],"argue":[81],"random":[84],"turn":[87],"skews":[88],"the":[89,121,129],"distributions":[90],"certain":[92],"segmentational":[93],"properties":[94],"(e.g.":[95],"token":[96],"lengths":[97],"and":[98],"amount":[99],"tokens":[101],"produced)":[102],"away":[103],"uniformity,":[105],"which":[106],"ends":[108],"up":[109],"hiding":[110],"meaningfully":[111],"diverse":[112],"tokenisations.":[113],"We":[114,164],"propose":[115],"alternative":[117],"uniform":[118],"sampler":[119],"using":[120],"same":[122],"graph,":[124],"but":[125],"weighted":[126],"by":[127],"counting":[128],"it.":[132],"Our":[133],"algorithm,":[135],"GRaMPa,":[136],"provides":[137],"hyperparameters":[138],"allowing":[139],"sampled":[140],"tokenisations":[141],"skew":[143],"towards":[144],"fewer,":[145],"longer":[146],"tokens.":[147],"Furthermore,":[148],"GRaMPa":[149,172],"single-pass,":[151],"guaranteeing":[152],"significantly":[153],"better":[154],"computational":[155],"complexity":[156],"than":[157],"previous":[158],"approaches":[159],"relying":[160],"on":[161,181],"sampling.":[163],"show":[165],"experimentally":[166],"trained":[170],"with":[171,189],"outperform":[173],"existing":[174],"regularising":[175],"data-scarce":[179],"setting":[180],"token-level":[182],"tasks":[183],"dependency":[186],"parsing,":[187],"especially":[188],"errors":[191],"present.":[192]},"counts_by_year":[],"updated_date":"2026-07-02T09:51:11.867554","created_date":"2025-10-10T00:00:00"}