{"id":"https://openalex.org/W2954507353","doi":"https://doi.org/10.21437/interspeech.2019-2811","title":"ShrinkML: End-to-End ASR Model Compression Using Reinforcement Learning","display_name":"ShrinkML: End-to-End ASR Model Compression Using Reinforcement Learning","publication_year":2019,"publication_date":"2019-09-13","ids":{"openalex":"https://openalex.org/W2954507353","doi":"https://doi.org/10.21437/interspeech.2019-2811","mag":"2954507353"},"language":"en","primary_location":{"id":"doi:10.21437/interspeech.2019-2811","is_oa":false,"landing_page_url":"https://doi.org/10.21437/interspeech.2019-2811","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Interspeech 2019","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/1907.03540","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5080541362","display_name":"\u0141ukasz Dudziak","orcid":"https://orcid.org/0000-0003-4929-265X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"\u0141ukasz Dudziak","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5010814272","display_name":"Mohamed S. Abdelfattah","orcid":"https://orcid.org/0000-0002-4568-8932"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Mohamed S. Abdelfattah","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5069973475","display_name":"Ravichander Vipperla","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ravichander Vipperla","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5049130978","display_name":"Stefanos Laskaridis","orcid":"https://orcid.org/0000-0002-8875-7328"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Stefanos Laskaridis","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5045638679","display_name":"Nicholas D. Lane","orcid":"https://orcid.org/0000-0002-2728-8273"},"institutions":[{"id":"https://openalex.org/I4210101778","display_name":"Samsung (United States)","ror":"https://ror.org/01bfbvm65","country_code":"US","type":"company","lineage":["https://openalex.org/I2250650973","https://openalex.org/I4210101778"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Nicholas D. Lane","raw_affiliation_strings":["[Samsung]"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"[Samsung]","institution_ids":["https://openalex.org/I4210101778"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":2,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"2235","last_page":"2239"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10201","display_name":"Speech Recognition and Synthesis","score":0.9995999932289124,"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/T10201","display_name":"Speech Recognition and Synthesis","score":0.9995999932289124,"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/T11309","display_name":"Music and Audio Processing","score":0.9987000226974487,"subfield":{"id":"https://openalex.org/subfields/1711","display_name":"Signal Processing"},"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/T10028","display_name":"Topic Modeling","score":0.9970999956130981,"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/reinforcement-learning","display_name":"Reinforcement learning","score":0.7772223949432373},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7274459600448608},{"id":"https://openalex.org/keywords/rank","display_name":"Rank (graph theory)","score":0.5295549035072327},{"id":"https://openalex.org/keywords/compression","display_name":"Compression (physics)","score":0.5240621566772461},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.5177899599075317},{"id":"https://openalex.org/keywords/singular-value-decomposition","display_name":"Singular value decomposition","score":0.46765583753585815},{"id":"https://openalex.org/keywords/set","display_name":"Set (abstract data type)","score":0.4531458020210266},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.36338961124420166},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.09197905659675598}],"concepts":[{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.7772223949432373},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7274459600448608},{"id":"https://openalex.org/C164226766","wikidata":"https://www.wikidata.org/wiki/Q7293202","display_name":"Rank (graph theory)","level":2,"score":0.5295549035072327},{"id":"https://openalex.org/C180016635","wikidata":"https://www.wikidata.org/wiki/Q2712821","display_name":"Compression (physics)","level":2,"score":0.5240621566772461},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5177899599075317},{"id":"https://openalex.org/C22789450","wikidata":"https://www.wikidata.org/wiki/Q420904","display_name":"Singular value decomposition","level":2,"score":0.46765583753585815},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.4531458020210266},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.36338961124420166},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.09197905659675598},{"id":"https://openalex.org/C114614502","wikidata":"https://www.wikidata.org/wiki/Q76592","display_name":"Combinatorics","level":1,"score":0.0},{"id":"https://openalex.org/C192562407","wikidata":"https://www.wikidata.org/wiki/Q228736","display_name":"Materials science","level":0,"score":0.0},{"id":"https://openalex.org/C159985019","wikidata":"https://www.wikidata.org/wiki/Q181790","display_name":"Composite material","level":1,"score":0.0},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.0}],"mesh":[],"locations_count":4,"locations":[{"id":"doi:10.21437/interspeech.2019-2811","is_oa":false,"landing_page_url":"https://doi.org/10.21437/interspeech.2019-2811","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Interspeech 2019","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:1907.03540","is_oa":true,"landing_page_url":"http://arxiv.org/abs/1907.03540","pdf_url":"https://arxiv.org/pdf/1907.03540","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"public-domain","license_id":"https://openalex.org/licenses/public-domain","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"text"},{"id":"mag:2954507353","is_oa":true,"landing_page_url":"https://arxiv.org/pdf/1907.03540","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"arXiv (Cornell University)","raw_type":null},{"id":"doi:10.48550/arxiv.1907.03540","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.1907.03540","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:1907.03540","is_oa":true,"landing_page_url":"http://arxiv.org/abs/1907.03540","pdf_url":"https://arxiv.org/pdf/1907.03540","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"public-domain","license_id":"https://openalex.org/licenses/public-domain","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"text"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/16","display_name":"Peace, Justice and strong institutions","score":0.7099999785423279}],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W2954507353.pdf","grobid_xml":"https://content.openalex.org/works/W2954507353.grobid-xml"},"referenced_works_count":23,"referenced_works":["https://openalex.org/W1494198834","https://openalex.org/W1586532344","https://openalex.org/W1902237438","https://openalex.org/W2056738732","https://openalex.org/W2294543795","https://openalex.org/W2327501763","https://openalex.org/W2507580616","https://openalex.org/W2535838896","https://openalex.org/W2546536770","https://openalex.org/W2748659049","https://openalex.org/W2799800213","https://openalex.org/W2799923439","https://openalex.org/W2886851211","https://openalex.org/W2888867175","https://openalex.org/W2889129739","https://openalex.org/W2905263927","https://openalex.org/W2962760690","https://openalex.org/W2962784628","https://openalex.org/W2963374479","https://openalex.org/W2964272710","https://openalex.org/W2964308564","https://openalex.org/W3007528493","https://openalex.org/W3008587939"],"related_works":["https://openalex.org/W2903933688","https://openalex.org/W3141188098","https://openalex.org/W3200231871","https://openalex.org/W2944890708","https://openalex.org/W3028274012","https://openalex.org/W2803180393","https://openalex.org/W3035825996","https://openalex.org/W2901006014","https://openalex.org/W3207441423","https://openalex.org/W3016791490","https://openalex.org/W3116063084","https://openalex.org/W3182057971","https://openalex.org/W3162026808","https://openalex.org/W3134587287","https://openalex.org/W3018492473","https://openalex.org/W186881209","https://openalex.org/W2949488962","https://openalex.org/W2606908067","https://openalex.org/W3201433989","https://openalex.org/W2292365397"],"abstract_inverted_index":{"End-to-end":[0],"automatic":[1],"speech":[2],"recognition":[3],"(ASR)":[4],"models":[5,166],"are":[6],"increasingly":[7],"large":[8],"and":[9,89,120,142],"complex":[10],"to":[11,30,38,124,145,168],"achieve":[12],"the":[13,32,61,76,79,82,90,113,132],"best":[14],"possible":[15],"accuracy.":[16],"In":[17],"this":[18],"paper,":[19],"we":[20,69,106,121,155],"build":[21],"an":[22,85,140],"AutoML":[23,67,118,159],"system":[24,160],"that":[25,130,157],"uses":[26],"reinforcement":[27],"learning":[28],"(RL)":[29],"optimize":[31],"per-layer":[33],"compression":[34,62,177],"ratios":[35],"when":[36],"applied":[37],"a":[39,93,147],"state-of-the-art":[40],"attention":[41],"based":[42],"end-to-end":[43],"ASR":[44,149],"model":[45,114],"composed":[46],"of":[47,78,84,92,112,134],"several":[48],"LSTM":[49],"layers.":[50],"We":[51],"use":[52],"singular":[53],"value":[54],"decomposition":[55],"(SVD)":[56],"low-rank":[57],"matrix":[58],"factorization":[59],"as":[60,75],"method.":[63],"For":[64],"our":[65,117,136,158],"RL-based":[66,137],"system,":[68,119],"focus":[70],"on":[71,110],"practical":[72,143],"considerations":[73],"such":[74],"choice":[77],"reward/punishment":[80],"functions,":[81],"formation":[83],"effective":[86,141],"search":[87,103,138],"space,":[88],"creation":[91],"representative":[94],"but":[95],"small":[96],"data":[97],"set":[98],"for":[99,175],"quick":[100],"evaluation":[101],"between":[102],"steps.":[104],"Finally,":[105],"present":[107],"accuracy":[108],"results":[109,128],"LibriSpeech":[111],"compressed":[115],"by":[116],"compare":[122],"it":[123],"manually-compressed":[125],"models.":[126],"Our":[127],"show":[129,156],"in":[131],"absence":[133],"retraining":[135,152],"is":[139,153],"method":[144],"compress":[146],"production-grade":[148],"system.":[150],"When":[151],"possible,":[154],"can":[161],"select":[162],"better":[163],"highly-compressed":[164],"seed":[165],"compared":[167],"manually":[169],"hand-crafted":[170],"rank":[171],"selection,":[172],"thus":[173],"allowing":[174],"more":[176],"than":[178],"previously":[179],"possible.":[180]},"counts_by_year":[{"year":2025,"cited_by_count":1},{"year":2024,"cited_by_count":1}],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2025-10-10T00:00:00"}