{"id":"https://openalex.org/W6892033892","doi":"https://doi.org/10.48550/arxiv.2505.16652","title":"Seeing Far and Clearly: Mitigating Hallucinations in MLLMs with Attention Causal Decoding","display_name":"Seeing Far and Clearly: Mitigating Hallucinations in MLLMs with Attention Causal Decoding","publication_year":2025,"publication_date":"2025-05-22","ids":{"openalex":"https://openalex.org/W6892033892","doi":"https://doi.org/10.48550/arxiv.2505.16652"},"language":"en","primary_location":{"id":"doi:10.48550/arxiv.2505.16652","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2505.16652","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":"article"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2505.16652","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Tang, Feilong","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Tang, Feilong","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Liu, Chengzhi","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Chengzhi","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Xu, Zhongxing","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xu, Zhongxing","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Hu, Ming","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hu, Ming","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Peng, Zelin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Peng, Zelin","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Yang, Zhiwei","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Zhiwei","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Su, Jionglong","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Su, Jionglong","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Lin, Minquan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lin, Minquan","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Peng, Yifan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Peng, Yifan","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Cheng, Xuelian","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Cheng, Xuelian","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Razzak, Imran","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Razzak, Imran","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":null,"display_name":"Ge, Zongyuan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ge, Zongyuan","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":12,"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":true,"primary_topic":{"id":"https://openalex.org/T13918","display_name":"Advanced Data and IoT Technologies","score":0.05959999933838844,"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/T13918","display_name":"Advanced Data and IoT Technologies","score":0.05959999933838844,"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/T14347","display_name":"Big Data and Digital Economy","score":0.04800000041723251,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/T14260","display_name":"Impact of AI and Big Data on Business and Society","score":0.042899999767541885,"subfield":{"id":"https://openalex.org/subfields/1803","display_name":"Management Science and Operations Research"},"field":{"id":"https://openalex.org/fields/18","display_name":"Decision Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.6833999752998352},{"id":"https://openalex.org/keywords/decoding-methods","display_name":"Decoding methods","score":0.6751000285148621},{"id":"https://openalex.org/keywords/outlier","display_name":"Outlier","score":0.5896000266075134},{"id":"https://openalex.org/keywords/encoding","display_name":"Encoding (memory)","score":0.49230000376701355},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.49050000309944153},{"id":"https://openalex.org/keywords/causal-reasoning","display_name":"Causal reasoning","score":0.46209999918937683},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.4602999985218048},{"id":"https://openalex.org/keywords/leverage","display_name":"Leverage (statistics)","score":0.43130001425743103},{"id":"https://openalex.org/keywords/masking","display_name":"Masking (illustration)","score":0.38370001316070557},{"id":"https://openalex.org/keywords/causal-model","display_name":"Causal model","score":0.3817000091075897},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.37139999866485596}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7127000093460083},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.6833999752998352},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.6751000285148621},{"id":"https://openalex.org/C79337645","wikidata":"https://www.wikidata.org/wiki/Q779824","display_name":"Outlier","level":2,"score":0.5896000266075134},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.49230000376701355},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.49050000309944153},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4772000014781952},{"id":"https://openalex.org/C115086926","wikidata":"https://www.wikidata.org/wiki/Q17004651","display_name":"Causal reasoning","level":3,"score":0.46209999918937683},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.4602999985218048},{"id":"https://openalex.org/C153083717","wikidata":"https://www.wikidata.org/wiki/Q6535263","display_name":"Leverage (statistics)","level":2,"score":0.43130001425743103},{"id":"https://openalex.org/C2777402240","wikidata":"https://www.wikidata.org/wiki/Q6783436","display_name":"Masking (illustration)","level":2,"score":0.38370001316070557},{"id":"https://openalex.org/C11671645","wikidata":"https://www.wikidata.org/wiki/Q5054567","display_name":"Causal model","level":2,"score":0.3817000091075897},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.37139999866485596},{"id":"https://openalex.org/C739882","wikidata":"https://www.wikidata.org/wiki/Q3560506","display_name":"Anomaly detection","level":2,"score":0.3644999861717224},{"id":"https://openalex.org/C180747234","wikidata":"https://www.wikidata.org/wiki/Q23373","display_name":"Cognitive psychology","level":1,"score":0.3447999954223633},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.33739998936653137},{"id":"https://openalex.org/C108154423","wikidata":"https://www.wikidata.org/wiki/Q1469792","display_name":"Salience (neuroscience)","level":2,"score":0.31470000743865967},{"id":"https://openalex.org/C2779458634","wikidata":"https://www.wikidata.org/wiki/Q24963715","display_name":"Debiasing","level":2,"score":0.31360000371932983},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.31029999256134033},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.30480000376701355},{"id":"https://openalex.org/C2779200073","wikidata":"https://www.wikidata.org/wiki/Q18395575","display_name":"Visual masking","level":4,"score":0.29980000853538513},{"id":"https://openalex.org/C2780186347","wikidata":"https://www.wikidata.org/wiki/Q11414","display_name":"Subnetwork","level":2,"score":0.28999999165534973},{"id":"https://openalex.org/C15744967","wikidata":"https://www.wikidata.org/wiki/Q9418","display_name":"Psychology","level":0,"score":0.289000004529953},{"id":"https://openalex.org/C2778112365","wikidata":"https://www.wikidata.org/wiki/Q3511065","display_name":"Sequence (biology)","level":2,"score":0.2840999960899353},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.2791999876499176},{"id":"https://openalex.org/C158600405","wikidata":"https://www.wikidata.org/wiki/Q5054566","display_name":"Causal inference","level":2,"score":0.2712000012397766},{"id":"https://openalex.org/C106399304","wikidata":"https://www.wikidata.org/wiki/Q868123","display_name":"Snowball sampling","level":2,"score":0.2526000142097473},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.24629999697208405},{"id":"https://openalex.org/C72010251","wikidata":"https://www.wikidata.org/wiki/Q5265688","display_name":"Determinantal point process","level":4,"score":0.24369999766349792},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.2425999939441681},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.2418999969959259},{"id":"https://openalex.org/C152948882","wikidata":"https://www.wikidata.org/wiki/Q4060686","display_name":"Belief propagation","level":3,"score":0.24079999327659607},{"id":"https://openalex.org/C97256817","wikidata":"https://www.wikidata.org/wiki/Q1462316","display_name":"Spurious relationship","level":2,"score":0.23899999260902405},{"id":"https://openalex.org/C2777508537","wikidata":"https://www.wikidata.org/wiki/Q7936620","display_name":"Visual reasoning","level":2,"score":0.23749999701976776},{"id":"https://openalex.org/C177774035","wikidata":"https://www.wikidata.org/wiki/Q1246948","display_name":"Granularity","level":2,"score":0.22840000689029694},{"id":"https://openalex.org/C114289077","wikidata":"https://www.wikidata.org/wiki/Q3284399","display_name":"Statistical model","level":2,"score":0.22599999606609344},{"id":"https://openalex.org/C169900460","wikidata":"https://www.wikidata.org/wiki/Q2200417","display_name":"Cognition","level":2,"score":0.22579999268054962},{"id":"https://openalex.org/C116834253","wikidata":"https://www.wikidata.org/wiki/Q2039217","display_name":"Identification (biology)","level":2,"score":0.22529999911785126},{"id":"https://openalex.org/C146249460","wikidata":"https://www.wikidata.org/wiki/Q2914991","display_name":"Fixation (population genetics)","level":3,"score":0.22519999742507935},{"id":"https://openalex.org/C61641136","wikidata":"https://www.wikidata.org/wiki/Q1107019","display_name":"Cognitive load","level":3,"score":0.218299999833107}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2505.16652","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2505.16652","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":"article"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2505.16652","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2505.16652","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":"article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Recent":[0],"advancements":[1],"in":[2,12,56,102],"multimodal":[3,71],"large":[4],"language":[5],"models":[6],"(MLLMs)":[7],"have":[8],"significantly":[9],"improved":[10],"performance":[11,206],"visual":[13],"question":[14],"answering.":[15],"However,":[16],"they":[17],"often":[18],"suffer":[19],"from":[20,47,129],"hallucinations.":[21,36],"In":[22],"this":[23,115],"work,":[24],"hallucinations":[25,33],"are":[26],"categorized":[27],"into":[28],"two":[29],"main":[30],"types:":[31],"initial":[32],"and":[34,93,213],"snowball":[35],"We":[37,147],"argue":[38],"that":[39,76],"adequate":[40],"contextual":[41,95],"information":[42,68],"can":[43],"be":[44],"extracted":[45],"directly":[46],"the":[48,57,84,103,135,154,159,186],"token":[49,145],"interaction":[50,78],"process.":[51],"Inspired":[52],"by":[53,106,133],"causal":[54,64,136,160],"inference":[55],"decoding":[58,123],"strategy,":[59],"we":[60,98,117],"propose":[61,99],"to":[62,66,86,100,110,125,165,169,188,190],"leverage":[63],"masks":[65],"establish":[67],"propagation":[69,104],"between":[70,79],"tokens.":[72,171],"The":[73,138],"hypothesis":[74],"is":[75,143,183],"insufficient":[77],"those":[80],"tokens":[81,109,131],"may":[82],"lead":[83],"model":[85,187],"rely":[87],"on":[88,210],"outlier":[89,108,130,170],"tokens,":[90,193],"overlooking":[91],"dense":[92],"rich":[94],"cues.":[96],"Therefore,":[97],"intervene":[101],"process":[105],"tackling":[107],"enhance":[111],"in-context":[112],"inference.":[113],"With":[114,199],"goal,":[116],"present":[118],"FarSight,":[119],"a":[120,173,179],"versatile":[121],"plug-and-play":[122],"strategy":[124],"reduce":[126],"attention":[127,150,164,167],"interference":[128],"merely":[132],"optimizing":[134],"mask.":[137],"heart":[139],"of":[140,158],"our":[141],"method":[142,177],"effective":[144],"propagation.":[146],"design":[148],"an":[149],"register":[151],"structure":[152],"within":[153],"upper":[155],"triangular":[156],"matrix":[157],"mask,":[161],"dynamically":[162],"allocating":[163],"capture":[166],"diverted":[168],"Moreover,":[172],"positional":[174],"awareness":[175],"encoding":[176],"with":[178],"diminishing":[180],"masking":[181],"rate":[182],"proposed,":[184],"allowing":[185],"attend":[189],"further":[191],"preceding":[192],"especially":[194],"for":[195],"video":[196,214],"sequence":[197],"tasks.":[198],"extensive":[200],"experiments,":[201],"FarSight":[202],"demonstrates":[203],"significant":[204],"hallucination-mitigating":[205],"across":[207],"different":[208],"MLLMs":[209],"both":[211],"image":[212],"benchmarks,":[215],"proving":[216],"its":[217],"effectiveness.":[218]},"counts_by_year":[],"updated_date":"2025-11-06T06:51:31.235846","created_date":"2025-10-10T00:00:00"}