{"id":"https://openalex.org/W7163185434","doi":"https://doi.org/10.48550/arxiv.2606.01666","title":"DOT-MoE: Differentiable Optimal Transport for MoEfication","display_name":"DOT-MoE: Differentiable Optimal Transport for MoEfication","publication_year":2026,"publication_date":"2026-06-01","ids":{"openalex":"https://openalex.org/W7163185434","doi":"https://doi.org/10.48550/arxiv.2606.01666"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.01666","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.01666","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2606.01666","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5137692259","display_name":"Udbhav Bamba","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bamba, Udbhav","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137664391","display_name":"Arnav Chavan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chavan, Arnav","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137655250","display_name":"Aryamaan Thakur","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Thakur, Aryamaan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137641575","display_name":"Steve Teig","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Teig, Steve","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5137643904","display_name":"Deepak Gupta","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gupta, Deepak","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":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.21320000290870667,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.21320000290870667,"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/T11704","display_name":"Mobile Crowdsensing and Crowdsourcing","score":0.1647000014781952,"subfield":{"id":"https://openalex.org/subfields/1706","display_name":"Computer Science Applications"},"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":0.0568000003695488,"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/inference","display_name":"Inference","score":0.5892999768257141},{"id":"https://openalex.org/keywords/differentiable-function","display_name":"Differentiable function","score":0.5777000188827515},{"id":"https://openalex.org/keywords/heuristic","display_name":"Heuristic","score":0.5541999936103821},{"id":"https://openalex.org/keywords/partition","display_name":"Partition (number theory)","score":0.4887000024318695},{"id":"https://openalex.org/keywords/estimator","display_name":"Estimator","score":0.4781000018119812},{"id":"https://openalex.org/keywords/decomposition","display_name":"Decomposition","score":0.414900004863739},{"id":"https://openalex.org/keywords/decoupling","display_name":"Decoupling (probability)","score":0.3885999917984009},{"id":"https://openalex.org/keywords/scaling","display_name":"Scaling","score":0.37790000438690186}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6141999959945679},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5892999768257141},{"id":"https://openalex.org/C202615002","wikidata":"https://www.wikidata.org/wiki/Q783507","display_name":"Differentiable function","level":2,"score":0.5777000188827515},{"id":"https://openalex.org/C173801870","wikidata":"https://www.wikidata.org/wiki/Q201413","display_name":"Heuristic","level":2,"score":0.5541999936103821},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.5479000210762024},{"id":"https://openalex.org/C42812","wikidata":"https://www.wikidata.org/wiki/Q1082910","display_name":"Partition (number theory)","level":2,"score":0.4887000024318695},{"id":"https://openalex.org/C185429906","wikidata":"https://www.wikidata.org/wiki/Q1130160","display_name":"Estimator","level":2,"score":0.4781000018119812},{"id":"https://openalex.org/C124681953","wikidata":"https://www.wikidata.org/wiki/Q339062","display_name":"Decomposition","level":2,"score":0.414900004863739},{"id":"https://openalex.org/C205606062","wikidata":"https://www.wikidata.org/wiki/Q5249645","display_name":"Decoupling (probability)","level":2,"score":0.3885999917984009},{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.37790000438690186},{"id":"https://openalex.org/C74172769","wikidata":"https://www.wikidata.org/wiki/Q1446839","display_name":"Routing (electronic design automation)","level":2,"score":0.3725999891757965},{"id":"https://openalex.org/C73555534","wikidata":"https://www.wikidata.org/wiki/Q622825","display_name":"Cluster analysis","level":2,"score":0.3628000020980835},{"id":"https://openalex.org/C105611402","wikidata":"https://www.wikidata.org/wiki/Q2976589","display_name":"Spectral clustering","level":3,"score":0.3402000069618225},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.33480000495910645},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.33070001006126404},{"id":"https://openalex.org/C22367795","wikidata":"https://www.wikidata.org/wiki/Q7625208","display_name":"Structured prediction","level":2,"score":0.313400000333786},{"id":"https://openalex.org/C2780186347","wikidata":"https://www.wikidata.org/wiki/Q11414","display_name":"Subnetwork","level":2,"score":0.3125},{"id":"https://openalex.org/C147764199","wikidata":"https://www.wikidata.org/wiki/Q6865248","display_name":"Minification","level":2,"score":0.2930999994277954},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.2799000144004822},{"id":"https://openalex.org/C127705205","wikidata":"https://www.wikidata.org/wiki/Q5748245","display_name":"Heuristics","level":2,"score":0.2759000062942505},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.2727000117301941},{"id":"https://openalex.org/C114809511","wikidata":"https://www.wikidata.org/wiki/Q1412924","display_name":"Flow network","level":2,"score":0.27149999141693115},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.26739999651908875},{"id":"https://openalex.org/C51485801","wikidata":"https://www.wikidata.org/wiki/Q16966861","display_name":"Efficient frontier","level":3,"score":0.2524000108242035}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.01666","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.01666","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2606.01666","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.01666","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"The":[0],"scaling":[1],"of":[2,21,45,91,102,164],"Large":[3],"Language":[4],"Models":[5],"(LLMs)":[6],"has":[7,52],"driven":[8],"significant":[9],"performance":[10,169],"gains":[11],"but":[12],"created":[13],"substantial":[14],"challenges":[15],"in":[16],"inference":[17,32],"efficiency.":[18],"While":[19],"Mixture":[20],"Experts":[22],"(MoEs)":[23],"architectures":[24,147],"address":[25],"this":[26,79],"by":[27,174],"decoupling":[28],"model":[29,106],"size":[30],"from":[31,36],"cost,":[33],"training":[34],"MoEs":[35,51],"scratch":[37],"is":[38],"often":[39],"unstable":[40],"and":[41,137,148,159],"compute":[42],"intensive.":[43],"Conversion":[44],"pre-trained":[46],"dense":[47,92,167],"models":[48],"into":[49,76],"sparse":[50],"emerged":[53],"as":[54,94,109],"an":[55],"alternative":[56],"solution;":[57],"however,":[58],"existing":[59],"methods":[60],"typically":[61],"rely":[62],"on":[63],"heuristic":[64,157],"neuron":[65,107],"clustering":[66],"or":[67],"random":[68],"splitting":[69],"to":[70,118,130],"partition":[71],"the":[72,89,133,138,165],"Feed-Forward":[73],"Network":[74],"(FFN)":[75],"experts.":[77],"In":[78],"work,":[80],"we":[81,105,125],"propose":[82],"DOT-MoE,":[83],"a":[84,95,110],"novel":[85],"framework":[86],"that":[87,151],"formulates":[88],"decomposition":[90],"layers":[93],"Differentiable":[96],"Optimal":[97],"Transport":[98],"(DOT)":[99],"problem.":[100],"Instead":[101],"static":[103],"heuristics,":[104],"assignment":[108,136],"balanced":[111],"transport":[112],"problem,":[113],"utilizing":[114],"differentiable":[115],"Sinkhorn-Knopp":[116],"iterations":[117],"enforce":[119],"strict":[120],"expert":[121],"capacity":[122],"constraints.":[123],"Furthermore,":[124],"utilize":[126],"Straight-Through":[127],"Estimators":[128],"(STE)":[129],"jointly":[131],"learn":[132],"discrete":[134],"neuron-to-expert":[135],"token-to-expert":[139],"routing":[140],"policy":[141],"end-to-end.":[142],"Extensive":[143],"experiments":[144],"across":[145],"multiple":[146],"benchmarks":[149],"demonstrate":[150],"DOT-MoE":[152],"significantly":[153],"outperforms":[154],"structured":[155],"pruning,":[156],"clustering,":[158],"random-split":[160],"baselines,":[161],"retaining":[162],"90%":[163],"original":[166],"model's":[168],"while":[170],"reducing":[171],"active":[172],"parameters":[173],"50%.":[175]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-06-03T00:00:00"}