{"id":"https://openalex.org/W4409158136","doi":"https://doi.org/10.1145/3690624.3709416","title":"HoME: Hierarchy of Multi-Gate Experts for Multi-Task Learning at Kuaishou","display_name":"HoME: Hierarchy of Multi-Gate Experts for Multi-Task Learning at Kuaishou","publication_year":2025,"publication_date":"2025-04-04","ids":{"openalex":"https://openalex.org/W4409158136","doi":"https://doi.org/10.1145/3690624.3709416"},"language":"en","primary_location":{"id":"doi:10.1145/3690624.3709416","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3690624.3709416","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.1","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://doi.org/10.1145/3690624.3709416","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5113390821","display_name":"Xu Wang","orcid":"https://orcid.org/0000-0003-0941-6232"},"institutions":[{"id":"https://openalex.org/I4401726859","display_name":"Kuaishou (China)","ror":"https://ror.org/0258as409","country_code":null,"type":"company","lineage":["https://openalex.org/I4401726859"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xu Wang","raw_affiliation_strings":["Kuaishou Technology, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0003-0941-6232","affiliations":[{"raw_affiliation_string":"Kuaishou Technology, Beijing, China","institution_ids":["https://openalex.org/I4401726859"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5017144784","display_name":"Jiangxia Cao","orcid":"https://orcid.org/0000-0003-2681-0119"},"institutions":[{"id":"https://openalex.org/I4401726859","display_name":"Kuaishou (China)","ror":"https://ror.org/0258as409","country_code":null,"type":"company","lineage":["https://openalex.org/I4401726859"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jiangxia Cao","raw_affiliation_strings":["Kuaishou Technology, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0003-2681-0119","affiliations":[{"raw_affiliation_string":"Kuaishou Technology, Beijing, China","institution_ids":["https://openalex.org/I4401726859"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5080130776","display_name":"Zhiyi Fu","orcid":"https://orcid.org/0000-0003-0260-6404"},"institutions":[{"id":"https://openalex.org/I4401726859","display_name":"Kuaishou (China)","ror":"https://ror.org/0258as409","country_code":null,"type":"company","lineage":["https://openalex.org/I4401726859"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhiyi Fu","raw_affiliation_strings":["Kuaishou Technology, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0003-0260-6404","affiliations":[{"raw_affiliation_string":"Kuaishou Technology, Beijing, China","institution_ids":["https://openalex.org/I4401726859"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5062939922","display_name":"Kun Gai","orcid":"https://orcid.org/0000-0002-3636-3618"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kun Gai","raw_affiliation_strings":["Unaffiliated, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0002-3636-3618","affiliations":[{"raw_affiliation_string":"Unaffiliated, Beijing, China","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5102865335","display_name":"Guorui Zhou","orcid":"https://orcid.org/0009-0002-8550-279X"},"institutions":[{"id":"https://openalex.org/I4401726859","display_name":"Kuaishou (China)","ror":"https://ror.org/0258as409","country_code":null,"type":"company","lineage":["https://openalex.org/I4401726859"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Guorui Zhou","raw_affiliation_strings":["Kuaishou Technology, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0002-8550-279X","affiliations":[{"raw_affiliation_string":"Kuaishou Technology, Beijing, China","institution_ids":["https://openalex.org/I4401726859"]}]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":11,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"2638","last_page":"2647"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12072","display_name":"Machine Learning and Algorithms","score":0.9772999882698059,"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/T12072","display_name":"Machine Learning and Algorithms","score":0.9772999882698059,"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/T10215","display_name":"Semantic Web and Ontologies","score":0.9527999758720398,"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/T13382","display_name":"Robotics and Automated Systems","score":0.9387999773025513,"subfield":{"id":"https://openalex.org/subfields/2207","display_name":"Control and Systems Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7326284646987915},{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.7087534666061401},{"id":"https://openalex.org/keywords/hierarchy","display_name":"Hierarchy","score":0.6458182334899902},{"id":"https://openalex.org/keywords/human\u2013computer-interaction","display_name":"Human\u2013computer interaction","score":0.38321858644485474},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.3660008907318115},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.3437942564487457},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.13194867968559265},{"id":"https://openalex.org/keywords/systems-engineering","display_name":"Systems engineering","score":0.059246569871902466}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7326284646987915},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.7087534666061401},{"id":"https://openalex.org/C31170391","wikidata":"https://www.wikidata.org/wiki/Q188619","display_name":"Hierarchy","level":2,"score":0.6458182334899902},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.38321858644485474},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3660008907318115},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3437942564487457},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.13194867968559265},{"id":"https://openalex.org/C201995342","wikidata":"https://www.wikidata.org/wiki/Q682496","display_name":"Systems engineering","level":1,"score":0.059246569871902466},{"id":"https://openalex.org/C34447519","wikidata":"https://www.wikidata.org/wiki/Q179522","display_name":"Market economy","level":1,"score":0.0},{"id":"https://openalex.org/C162324750","wikidata":"https://www.wikidata.org/wiki/Q8134","display_name":"Economics","level":0,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3690624.3709416","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3690624.3709416","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.1","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.1145/3690624.3709416","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3690624.3709416","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.1","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":16,"referenced_works":["https://openalex.org/W398859631","https://openalex.org/W2117130368","https://openalex.org/W2510317721","https://openalex.org/W2517190429","https://openalex.org/W2586505867","https://openalex.org/W2962989965","https://openalex.org/W2963168538","https://openalex.org/W2973172293","https://openalex.org/W3087931390","https://openalex.org/W3141797743","https://openalex.org/W3209943551","https://openalex.org/W4385562613","https://openalex.org/W4392979802","https://openalex.org/W4402671950","https://openalex.org/W6600195515","https://openalex.org/W6820618299"],"related_works":["https://openalex.org/W2961085424","https://openalex.org/W4306674287","https://openalex.org/W4387369504","https://openalex.org/W4394896187","https://openalex.org/W3170094116","https://openalex.org/W4386462264","https://openalex.org/W3107602296","https://openalex.org/W4364306694","https://openalex.org/W4312192474","https://openalex.org/W4283697347"],"abstract_inverted_index":{"In":[0,17,145],"this":[1],"paper,":[2],"we":[3,55,119,148,160],"present":[4],"the":[5,9,24,50,107,183],"practical":[6],"problems":[7],"and":[8,33,39,80,172,190],"lessons":[10],"learned":[11],"at":[12],"short-video":[13],"services":[14],"from":[15,192],"Kuaishou.":[16],"industry,":[18],"a":[19],"widely-used":[20],"multi-task":[21],"framework":[22],"is":[23],"Mixture-of-Experts":[25],"(MoE)":[26],"paradigm,":[27],"which":[28,130],"always":[29],"introduces":[30],"some":[31,81,122,140,163],"shared":[32],"specific":[34],"experts":[35,82],"for":[36,93,114],"each":[37],"task":[38],"then":[40],"uses":[41],"gate":[42,94],"networks":[43,95],"to":[44,96,100,110,156,168,176,202],"measure":[45],"related":[46],"experts'":[47,74],"contributions.":[48],"Although":[49],"MoE":[51],"achieves":[52],"remarkable":[53],"improvements,":[54],"still":[56],"observe":[57],"three":[58],"anomalies":[59],"that":[60,73,121,132,154,162,182],"seriously":[61],"affect":[62],"model":[63],"performances":[64],"in":[65],"our":[66,146],"iteration:":[67],"(1)":[68],"Expert":[69,104,143],"Collapse:":[70],"We":[71],"found":[72],"output":[75],"distributions":[76],"are":[77,124],"significantly":[78],"different,":[79],"have":[83,149],"over":[84],"90%":[85],"zero":[86],"activations":[87],"with":[88],"ReLU,":[89],"making":[90],"it":[91],"hard":[92],"assign":[97,173],"fair":[98],"weights":[99,175],"balance":[101],"experts.":[102],"(2)":[103],"Degradation:":[105],"Ideally,":[106],"shared-expert":[108],"aims":[109],"provide":[111],"predictive":[112],"information":[113],"all":[115],"tasks":[116,153,166],"simultaneously.":[117],"Nevertheless,":[118],"find":[120,161],"shared-experts":[123,133,184],"occupied":[125],"by":[126],"only":[127],"one":[128],"task,":[129],"indicates":[131],"lost":[134],"their":[135,170,203],"ability":[136],"but":[137,159],"degenerated":[138],"into":[139,199],"specific-experts.":[141],"(3)":[142],"Underfitting:":[144],"services,":[147],"dozens":[150],"of":[151],"behavior":[152],"need":[155],"be":[157,181],"predicted,":[158],"data-sparse":[164],"prediction":[165],"tend":[167],"ignore":[169],"specific-experts":[171,196],"large":[174],"shared-experts.":[177],"The":[178],"reason":[179],"might":[180],"can":[185],"perceive":[186],"more":[187],"gradient":[188],"updates":[189],"knowledge":[191],"dense":[193],"tasks,":[194],"while":[195],"easily":[197],"fall":[198],"underfitting":[200],"due":[201],"sparse":[204],"behaviors.":[205]},"counts_by_year":[{"year":2026,"cited_by_count":7},{"year":2025,"cited_by_count":4}],"updated_date":"2026-07-21T08:15:58.654021","created_date":"2025-10-10T00:00:00"}