{"id":"https://openalex.org/W4416037068","doi":"https://doi.org/10.18653/v1/2025.emnlp-main.118","title":"Integral Transformer: Denoising Attention, Not Too Much Not Too Little","display_name":"Integral Transformer: Denoising Attention, Not Too Much Not Too Little","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4416037068","doi":"https://doi.org/10.18653/v1/2025.emnlp-main.118"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2025.emnlp-main.118","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.emnlp-main.118","pdf_url":"https://aclanthology.org/2025.emnlp-main.118.pdf","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 2025 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.emnlp-main.118.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5008923192","display_name":"Ivan Kobyzev","orcid":"https://orcid.org/0000-0003-1934-4842"},"institutions":[{"id":"https://openalex.org/I4210159102","display_name":"Huawei Technologies (Sweden)","ror":"https://ror.org/0500fyd17","country_code":"SE","type":"company","lineage":["https://openalex.org/I2250955327","https://openalex.org/I4210159102"]}],"countries":["SE"],"is_corresponding":true,"raw_author_name":"Ivan Kobyzev","raw_affiliation_strings":["Huawei Noah's Ark Lab , Montreal Research Center , Canada"],"affiliations":[{"raw_affiliation_string":"Huawei Noah's Ark Lab , Montreal Research Center , Canada","institution_ids":["https://openalex.org/I4210159102"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5044793533","display_name":"Abbas Ghaddar","orcid":null},"institutions":[{"id":"https://openalex.org/I4210159102","display_name":"Huawei Technologies (Sweden)","ror":"https://ror.org/0500fyd17","country_code":"SE","type":"company","lineage":["https://openalex.org/I2250955327","https://openalex.org/I4210159102"]}],"countries":["SE"],"is_corresponding":false,"raw_author_name":"Abbas Ghaddar","raw_affiliation_strings":["Huawei Noah's Ark Lab , Montreal Research Center , Canada"],"affiliations":[{"raw_affiliation_string":"Huawei Noah's Ark Lab , Montreal Research Center , Canada","institution_ids":["https://openalex.org/I4210159102"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101673967","display_name":"Daoyu Hu","orcid":"https://orcid.org/0000-0002-8990-9581"},"institutions":[{"id":"https://openalex.org/I4210159102","display_name":"Huawei Technologies (Sweden)","ror":"https://ror.org/0500fyd17","country_code":"SE","type":"company","lineage":["https://openalex.org/I2250955327","https://openalex.org/I4210159102"]}],"countries":["SE"],"is_corresponding":false,"raw_author_name":"Dingtao Hu","raw_affiliation_strings":["Huawei Noah's Ark Lab , Montreal Research Center , Canada"],"affiliations":[{"raw_affiliation_string":"Huawei Noah's Ark Lab , Montreal Research Center , Canada","institution_ids":["https://openalex.org/I4210159102"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5019715118","display_name":"Boxing Chen","orcid":"https://orcid.org/0000-0002-3170-4858"},"institutions":[{"id":"https://openalex.org/I4210159102","display_name":"Huawei Technologies (Sweden)","ror":"https://ror.org/0500fyd17","country_code":"SE","type":"company","lineage":["https://openalex.org/I2250955327","https://openalex.org/I4210159102"]}],"countries":["SE"],"is_corresponding":false,"raw_author_name":"Boxing Chen","raw_affiliation_strings":["Huawei Noah's Ark Lab , Montreal Research Center , Canada"],"affiliations":[{"raw_affiliation_string":"Huawei Noah's Ark Lab , Montreal Research Center , Canada","institution_ids":["https://openalex.org/I4210159102"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":4,"corresponding_author_ids":["https://openalex.org/A5008923192"],"corresponding_institution_ids":["https://openalex.org/I4210159102"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.34621709,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"2337","last_page":"2354"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12300","display_name":"Advanced Electrical Measurement Techniques","score":0.03460000082850456,"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"}},"topics":[{"id":"https://openalex.org/T12300","display_name":"Advanced Electrical Measurement Techniques","score":0.03460000082850456,"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/T11034","display_name":"Digital Filter Design and Implementation","score":0.033900000154972076,"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/T11021","display_name":"ECG Monitoring and Analysis","score":0.019300000742077827,"subfield":{"id":"https://openalex.org/subfields/2705","display_name":"Cardiology and Cardiovascular Medicine"},"field":{"id":"https://openalex.org/fields/27","display_name":"Medicine"},"domain":{"id":"https://openalex.org/domains/4","display_name":"Health Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/noise","display_name":"Noise (video)","score":0.3781000077724457},{"id":"https://openalex.org/keywords/noise-reduction","display_name":"Noise reduction","score":0.35530000925064087},{"id":"https://openalex.org/keywords/measure","display_name":"Measure (data warehouse)","score":0.2619999945163727},{"id":"https://openalex.org/keywords/pattern-recognition","display_name":"Pattern recognition (psychology)","score":0.25859999656677246},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.226500004529953}],"concepts":[{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.4018999934196472},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.38100001215934753},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.3781000077724457},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.37139999866485596},{"id":"https://openalex.org/C163294075","wikidata":"https://www.wikidata.org/wiki/Q581861","display_name":"Noise reduction","level":2,"score":0.35530000925064087},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.263700008392334},{"id":"https://openalex.org/C2780009758","wikidata":"https://www.wikidata.org/wiki/Q6804172","display_name":"Measure (data warehouse)","level":2,"score":0.2619999945163727},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.25859999656677246},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.226500004529953},{"id":"https://openalex.org/C14036430","wikidata":"https://www.wikidata.org/wiki/Q3736076","display_name":"Function (biology)","level":2,"score":0.2152000069618225}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.emnlp-main.118","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.emnlp-main.118","pdf_url":"https://aclanthology.org/2025.emnlp-main.118.pdf","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 2025 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.emnlp-main.118","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.emnlp-main.118","pdf_url":"https://aclanthology.org/2025.emnlp-main.118.pdf","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 2025 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4416037068.pdf","grobid_xml":"https://content.openalex.org/works/W4416037068.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Softmax":[0],"self-attention":[1,105],"often":[2],"assigns":[3],"disproportionate":[4],"weight":[5],"to":[6],"semantically":[7],"uninformative":[8],"tokens":[9,13,75],"such":[10],"as":[11,19],"special":[12,74],"and":[14,27,88,95,113,122],"punctuation,":[15],"a":[16,51],"phenomenon":[17],"known":[18],"attention":[20,37,57,90,120],"noise.While":[21],"recent":[22],"methods":[23],"like":[24],"Cog":[25],"Attention":[26],"the":[28,48,63,71,107,115],"Differential":[29,89],"Transformer":[30,109,117],"have":[31],"addressed":[32],"this":[33,44],"by":[34,58],"introducing":[35],"negative":[36],"scores,":[38],"they":[39],"risk":[40],"discarding":[41],"useful":[42],"information.In":[43],"paper,":[45],"we":[46],"propose":[47],"Integral":[49,116],"Transformer,":[50],"novel":[52],"selfattention":[53],"mechanism":[54],"that":[55,82,102,114],"denoises":[56],"integrating":[59],"signals":[60],"sampled":[61],"from":[62],"logit":[64],"distribution.Our":[65],"approach":[66],"mitigates":[67],"noise":[68],"while":[69],"preserving":[70],"contributions":[72],"of":[73,132],"critical":[76],"for":[77],"model":[78,84],"performance.Extensive":[79],"experiments":[80],"demonstrate":[81],"our":[83,99],"outperforms":[85],"vanilla,":[86],"Cog,":[87],"variants":[91],"on":[92],"well-established":[93],"knowledge":[94],"reasoning":[96],"language":[97],"benchmarks.Moreover,":[98],"analysis":[100,131],"reveals":[101],"employing":[103],"vanilla":[104],"in":[106,126],"lower":[108],"layers":[110],"enhances":[111],"performance":[112],"effectively":[118],"balances":[119],"distributions":[121],"reduces":[123],"rank":[124],"collapse":[125],"upper":[127],"layers.look":[128],"at?":[129],"an":[130],"BERT's":[133],"attention.In":[134]},"counts_by_year":[],"updated_date":"2026-03-12T06:13:28.667946","created_date":"2025-11-08T00:00:00"}