{"id":"https://openalex.org/W7134052986","doi":"https://doi.org/10.48550/arxiv.2603.05048","title":"MCEL: Margin-Based Cross-Entropy Loss for Error-Tolerant Quantized Neural Networks","display_name":"MCEL: Margin-Based Cross-Entropy Loss for Error-Tolerant Quantized Neural Networks","publication_year":2026,"publication_date":"2026-03-05","ids":{"openalex":"https://openalex.org/W7134052986","doi":"https://doi.org/10.48550/arxiv.2603.05048"},"language":null,"primary_location":{"id":"pmh:doi:10.48550/arxiv.2603.05048","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","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":"publisher-specific-oa","license_id":"https://openalex.org/licenses/publisher-specific-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},"type":"article","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":null,"any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5025566655","display_name":"Mikail Yayla","orcid":"https://orcid.org/0000-0002-4134-952X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yayla, Mikail","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5128226046","display_name":"Akash Kumar","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kumar, Akash","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":0.26251413,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"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/T10036","display_name":"Advanced Neural Network Applications","score":0.39469999074935913,"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"}},"topics":[{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.39469999074935913,"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"}},{"id":"https://openalex.org/T10502","display_name":"Advanced Memory and Neural Computing","score":0.18119999766349792,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11005","display_name":"Radiation Effects in Electronics","score":0.07339999824762344,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic 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/robustness","display_name":"Robustness (evolution)","score":0.7408999800682068},{"id":"https://openalex.org/keywords/quantization","display_name":"Quantization (signal processing)","score":0.5523999929428101},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.5318999886512756},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.5246999859809875},{"id":"https://openalex.org/keywords/margin","display_name":"Margin (machine learning)","score":0.4796000123023987},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.43970000743865967},{"id":"https://openalex.org/keywords/error-detection-and-correction","display_name":"Error detection and correction","score":0.43220001459121704},{"id":"https://openalex.org/keywords/word-error-rate","display_name":"Word error rate","score":0.35690000653266907}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7484999895095825},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.7408999800682068},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.5523999929428101},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.5318999886512756},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.5246999859809875},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.5149999856948853},{"id":"https://openalex.org/C774472","wikidata":"https://www.wikidata.org/wiki/Q6760393","display_name":"Margin (machine learning)","level":2,"score":0.4796000123023987},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.43970000743865967},{"id":"https://openalex.org/C103088060","wikidata":"https://www.wikidata.org/wiki/Q1062839","display_name":"Error detection and correction","level":2,"score":0.43220001459121704},{"id":"https://openalex.org/C40969351","wikidata":"https://www.wikidata.org/wiki/Q3516228","display_name":"Word error rate","level":2,"score":0.35690000653266907},{"id":"https://openalex.org/C56296756","wikidata":"https://www.wikidata.org/wiki/Q840922","display_name":"Bit error rate","level":3,"score":0.32850000262260437},{"id":"https://openalex.org/C167085575","wikidata":"https://www.wikidata.org/wiki/Q6803654","display_name":"Mean squared prediction error","level":2,"score":0.31709998846054077},{"id":"https://openalex.org/C204323151","wikidata":"https://www.wikidata.org/wiki/Q905424","display_name":"Range (aeronautics)","level":2,"score":0.3125},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.3052000105381012},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.288100004196167},{"id":"https://openalex.org/C2984842247","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep neural networks","level":3,"score":0.28380000591278076},{"id":"https://openalex.org/C139945424","wikidata":"https://www.wikidata.org/wiki/Q1940696","display_name":"Mean squared error","level":2,"score":0.27469998598098755},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.2732999920845032},{"id":"https://openalex.org/C2780586882","wikidata":"https://www.wikidata.org/wiki/Q7520643","display_name":"Simple (philosophy)","level":2,"score":0.2703999876976013},{"id":"https://openalex.org/C63540848","wikidata":"https://www.wikidata.org/wiki/Q3140932","display_name":"Fault tolerance","level":2,"score":0.2590999901294708}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:doi:10.48550/arxiv.2603.05048","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","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":"publisher-specific-oa","license_id":"https://openalex.org/licenses/publisher-specific-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},{"id":"doi:10.48550/arxiv.2603.05048","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.05048","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:doi:10.48550/arxiv.2603.05048","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","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":"publisher-specific-oa","license_id":"https://openalex.org/licenses/publisher-specific-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},"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":{"Robustness":[0],"to":[1,28,42,106,173,207,225,246],"bit":[2,30,37,53,108,121,203,248],"errors":[3,109],"is":[4,35,223],"a":[5,43,117,137,177,193,233,241],"key":[6],"requirement":[7],"for":[8,71,84,212,236],"the":[9,101,128,141,154,159,255],"reliable":[10],"use":[11],"of":[12,158,186,195,216,257],"neural":[13],"networks":[14],"(NNs)":[15],"on":[16,87,112,132,265],"emerging":[17],"approximate":[18,89,266],"computing":[19,90,267],"platforms":[20,91],"and":[21,68,92,124,192,228,243,260,268],"error-prone":[22,93],"memory":[23,94,269],"technologies.":[24,95],"A":[25],"common":[26],"approach":[27],"achieve":[29],"error":[31,45,66,78,122,204,214],"tolerance":[32,123],"in":[33,49,176,210],"NNs":[34,105],"injecting":[36],"flips":[38],"during":[39],"training":[40],"according":[41],"predefined":[44],"model.":[46],"While":[47],"effective":[48],"certain":[50],"scenarios,":[51],"training-time":[52,247],"flip":[54,249],"injection":[55,79],"introduces":[56,165],"substantial":[57],"computational":[58],"overhead,":[59],"often":[60],"degrades":[61],"inference":[62],"accuracy":[63,211],"at":[64,127],"high":[65],"rates,":[67],"scales":[69],"poorly":[70],"larger":[72],"NN":[73,190,258],"architectures.":[74],"These":[75],"limitations":[76],"make":[77],"an":[80,166,213],"increasingly":[81],"impractical":[82],"solution":[83],"ensuring":[85],"robustness":[86,172,259],"future":[88],"In":[96],"this":[97,133],"work,":[98],"we":[99,135],"investigate":[100],"mechanisms":[102],"that":[103,170,199],"enable":[104],"tolerate":[107],"without":[110],"relying":[111],"error-aware":[113],"training.":[114],"We":[115],"establish":[116],"direct":[118],"connection":[119],"between":[120],"classification":[125],"margins":[126],"output":[129],"layer.":[130],"Building":[131],"insight,":[134],"propose":[136],"novel":[138],"loss":[139],"function,":[140],"Margin":[142],"Cross-Entropy":[143],"Loss":[144],"(MCEL),":[145],"which":[146],"explicitly":[147],"promotes":[148],"logit-level":[149],"margin":[150,168],"separation":[151],"while":[152],"preserving":[153],"favorable":[155],"optimization":[156],"properties":[157],"standard":[160,237],"cross-entropy":[161],"loss.":[162],"Furthermore,":[163],"MCEL":[164,200,221],"interpretable":[167],"parameter":[169],"allows":[171],"be":[174,230],"tuned":[175],"principled":[178,244],"manner.":[179],"Extensive":[180],"experimental":[181],"evaluations":[182],"across":[183],"multiple":[184],"datasets":[185],"varying":[187],"complexity,":[188],"diverse":[189],"architectures,":[191],"range":[194],"quantization":[196],"schemes":[197],"demonstrate":[198],"substantially":[201],"improves":[202],"tolerance,":[205],"up":[206],"15":[208],"%":[209],"rate":[215],"1":[217],"%.":[218],"Our":[219],"proposed":[220],"method":[222],"simple":[224],"implement,":[226],"efficient,":[227],"can":[229],"integrated":[231],"as":[232],"drop-in":[234],"replacement":[235],"CEL.":[238],"It":[239],"provides":[240],"scalable":[242],"alternative":[245],"injection,":[250],"offering":[251],"new":[252],"insights":[253],"into":[254],"origins":[256],"enabling":[261],"more":[262],"efficient":[263],"deployment":[264],"systems.":[270]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-03-07T00:00:00"}