{"id":"https://openalex.org/W4415708531","doi":"https://doi.org/10.1109/icme59968.2025.11208943","title":"Advancing Safe Language Generation: Exploring Alternative Constrained RLHF","display_name":"Advancing Safe Language Generation: Exploring Alternative Constrained RLHF","publication_year":2025,"publication_date":"2025-06-30","ids":{"openalex":"https://openalex.org/W4415708531","doi":"https://doi.org/10.1109/icme59968.2025.11208943"},"language":null,"primary_location":{"id":"doi:10.1109/icme59968.2025.11208943","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icme59968.2025.11208943","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 IEEE International Conference on Multimedia and Expo (ICME)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5100513053","display_name":"Fanyu Meng","orcid":"https://orcid.org/0009-0000-7348-4147"},"institutions":[{"id":"https://openalex.org/I180662265","display_name":"China Mobile (China)","ror":"https://ror.org/05gftfe97","country_code":"CN","type":"company","lineage":["https://openalex.org/I180662265"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Fanyu Meng","raw_affiliation_strings":["China Mobile Research Institute,Beijing,China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"China Mobile Research Institute,Beijing,China","institution_ids":["https://openalex.org/I180662265"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Zhixin Bai","orcid":null},"institutions":[{"id":"https://openalex.org/I881766915","display_name":"Nanjing University","ror":"https://ror.org/01rxvg760","country_code":"CN","type":"education","lineage":["https://openalex.org/I881766915"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhixin Bai","raw_affiliation_strings":["Nanjing University,State Key Laboratory for Novel Software Technology,Nanjing,China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Nanjing University,State Key Laboratory for Novel Software Technology,Nanjing,China","institution_ids":["https://openalex.org/I881766915"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Yanming Wang","orcid":null},"institutions":[{"id":"https://openalex.org/I881766915","display_name":"Nanjing University","ror":"https://ror.org/01rxvg760","country_code":"CN","type":"education","lineage":["https://openalex.org/I881766915"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yanming Wang","raw_affiliation_strings":["Nanjing University,State Key Laboratory for Novel Software Technology,Nanjing,China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Nanjing University,State Key Laboratory for Novel Software Technology,Nanjing,China","institution_ids":["https://openalex.org/I881766915"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5066091819","display_name":"Jing Huo","orcid":"https://orcid.org/0000-0002-8504-455X"},"institutions":[{"id":"https://openalex.org/I881766915","display_name":"Nanjing University","ror":"https://ror.org/01rxvg760","country_code":"CN","type":"education","lineage":["https://openalex.org/I881766915"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jing Huo","raw_affiliation_strings":["Nanjing University,State Key Laboratory for Novel Software Technology,Nanjing,China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Nanjing University,State Key Laboratory for Novel Software Technology,Nanjing,China","institution_ids":["https://openalex.org/I881766915"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100736483","display_name":"Boyan Wang","orcid":"https://orcid.org/0000-0003-0088-2106"},"institutions":[{"id":"https://openalex.org/I308837","display_name":"Suzhou University of Science and Technology","ror":"https://ror.org/04en8wb91","country_code":"CN","type":"education","lineage":["https://openalex.org/I308837"]},{"id":"https://openalex.org/I36399199","display_name":"Nanjing University of Science and Technology","ror":"https://ror.org/00xp9wg62","country_code":"CN","type":"education","lineage":["https://openalex.org/I36399199"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Boyan Wang","raw_affiliation_strings":["Nanjing University,School of Intelligence Science and Technology,Suzhou,China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Nanjing University,School of Intelligence Science and Technology,Suzhou,China","institution_ids":["https://openalex.org/I308837","https://openalex.org/I36399199"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100613753","display_name":"Xi Yang","orcid":"https://orcid.org/0000-0002-6828-7744"},"institutions":[{"id":"https://openalex.org/I180662265","display_name":"China Mobile (China)","ror":"https://ror.org/05gftfe97","country_code":"CN","type":"company","lineage":["https://openalex.org/I180662265"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xi Yang","raw_affiliation_strings":["China Mobile (Suzhou) Software Technology Co., Ltd.,Suzhou,China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"China Mobile (Suzhou) Software Technology Co., Ltd.,Suzhou,China","institution_ids":["https://openalex.org/I180662265"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5038207720","display_name":"Yang Gao","orcid":"https://orcid.org/0000-0002-2150-5986"},"institutions":[{"id":"https://openalex.org/I881766915","display_name":"Nanjing University","ror":"https://ror.org/01rxvg760","country_code":"CN","type":"education","lineage":["https://openalex.org/I881766915"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yang Gao","raw_affiliation_strings":["Nanjing University,State Key Laboratory for Novel Software Technology,Nanjing,China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Nanjing University,State Key Laboratory for Novel Software Technology,Nanjing,China","institution_ids":["https://openalex.org/I881766915"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":4,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"6"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.6388999819755554,"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/T10028","display_name":"Topic Modeling","score":0.6388999819755554,"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/T12262","display_name":"Hate Speech and Cyberbullying Detection","score":0.06599999964237213,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.04039999842643738,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/helpfulness","display_name":"Helpfulness","score":0.8659999966621399},{"id":"https://openalex.org/keywords/compromise","display_name":"Compromise","score":0.8064000010490417},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.6448000073432922},{"id":"https://openalex.org/keywords/quality","display_name":"Quality (philosophy)","score":0.5300999879837036},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.4887000024318695},{"id":"https://openalex.org/keywords/variety","display_name":"Variety (cybernetics)","score":0.3084000051021576}],"concepts":[{"id":"https://openalex.org/C2781265381","wikidata":"https://www.wikidata.org/wiki/Q5710255","display_name":"Helpfulness","level":2,"score":0.8659999966621399},{"id":"https://openalex.org/C46355384","wikidata":"https://www.wikidata.org/wiki/Q726686","display_name":"Compromise","level":2,"score":0.8064000010490417},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7462000250816345},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.6448000073432922},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.5300999879837036},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.4887000024318695},{"id":"https://openalex.org/C112930515","wikidata":"https://www.wikidata.org/wiki/Q4389547","display_name":"Risk analysis (engineering)","level":1,"score":0.484499990940094},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.35199999809265137},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.31790000200271606},{"id":"https://openalex.org/C136197465","wikidata":"https://www.wikidata.org/wiki/Q1729295","display_name":"Variety (cybernetics)","level":2,"score":0.3084000051021576},{"id":"https://openalex.org/C168031717","wikidata":"https://www.wikidata.org/wiki/Q1530280","display_name":"Balance (ability)","level":2,"score":0.3082999885082245},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.3018999993801117},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.2872999906539917},{"id":"https://openalex.org/C179603123","wikidata":"https://www.wikidata.org/wiki/Q1941921","display_name":"Modeling language","level":3,"score":0.28529998660087585},{"id":"https://openalex.org/C539667460","wikidata":"https://www.wikidata.org/wiki/Q2414942","display_name":"Management science","level":1,"score":0.2800999879837036},{"id":"https://openalex.org/C2775924081","wikidata":"https://www.wikidata.org/wiki/Q55608371","display_name":"Control (management)","level":2,"score":0.25940001010894775},{"id":"https://openalex.org/C184356942","wikidata":"https://www.wikidata.org/wiki/Q830382","display_name":"Best practice","level":2,"score":0.2542000114917755}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/icme59968.2025.11208943","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icme59968.2025.11208943","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 IEEE International Conference on Multimedia and Expo (ICME)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"}],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":11,"referenced_works":["https://openalex.org/W2964223283","https://openalex.org/W2996728628","https://openalex.org/W3034850762","https://openalex.org/W3201174429","https://openalex.org/W3207316473","https://openalex.org/W4285210452","https://openalex.org/W4285600796","https://openalex.org/W4385570452","https://openalex.org/W4399917664","https://openalex.org/W4401042581","https://openalex.org/W4401248175"],"related_works":[],"abstract_inverted_index":{"As":[0],"Large":[1],"Language":[2],"Models":[3],"(LLMs)":[4],"have":[5,59,128],"been":[6],"increasingly":[7],"deployed":[8],"for":[9,163],"various":[10,133],"language":[11,96,164,179],"generation":[12,180],"tasks,":[13,136],"ensuring":[14],"the":[15,70,83,113,141,146,156,170,183],"safety":[16,53,67,149],"and":[17,39,54,117,125,137,148],"appropriateness":[18],"of":[19,85,115,150,158,185],"generated":[20,151],"content":[21,40],"has":[22],"become":[23],"a":[24],"critical":[25],"challenge.":[26],"While":[27],"these":[28,79],"models":[29],"excel":[30],"at":[31],"tasks":[32],"such":[33],"as":[34],"dialogue":[35],"generation,":[36],"text":[37],"completion,":[38],"creation,":[41],"they":[42],"may":[43],"inadvertently":[44],"generate":[45],"harmful":[46],"or":[47],"inappropriate":[48],"responses":[49],"that":[50,110,169],"compromise":[51,88],"user":[52],"model":[55,186],"reliability.":[56],"Recent":[57],"advancements":[58],"attempted":[60],"to":[61,91,144,176],"address":[62,99],"this":[63,100],"problem":[64],"by":[65],"incorporating":[66],"constraints":[68],"within":[69],"Reinforcement":[71],"Learning":[72],"with":[73,82],"Human":[74],"Feedback":[75],"(RLHF)":[76],"framework.":[77],"However,":[78],"approaches":[80],"struggle":[81],"complexity":[84],"selecting":[86],"appropriate":[87],"parameters,":[89],"leading":[90],"suboptimal":[92],"performance":[93,131,157],"in":[94,132],"safe":[95,178],"generation.":[97,165],"To":[98],"problem,":[101],"we":[102,122,154],"explore":[103],"advanced":[104],"alternative":[105],"first-order":[106],"constrained":[107,160],"optimization":[108,161],"strategies":[109],"dynamically":[111],"adjust":[112],"balance":[114,145],"helpfulness":[116,184],"harmlessness":[118],"during":[119],"training.":[120],"Specifically,":[121],"choose":[123],"FOCOPS":[124],"P3O,":[126],"which":[127],"demonstrated":[129],"strong":[130],"reinforcement":[134],"learning":[135],"integrate":[138],"them":[139],"into":[140],"RLHF":[142],"framework":[143],"quality":[147],"responses.":[152,187],"Then":[153],"evaluate":[155],"different":[159],"methods":[162],"The":[166],"results":[167],"indicate":[168],"P3O":[171],"algorithm":[172],"shows":[173],"significant":[174],"potential":[175],"advance":[177],"while":[181],"maintaining":[182]},"counts_by_year":[],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-30T00:00:00"}
