{"id":"https://openalex.org/W4416420249","doi":"https://doi.org/10.1016/j.patcog.2025.112765","title":"Visual reasoning consistency and robustness analysis of multimodal LLMs","display_name":"Visual reasoning consistency and robustness analysis of multimodal LLMs","publication_year":2025,"publication_date":"2025-11-20","ids":{"openalex":"https://openalex.org/W4416420249","doi":"https://doi.org/10.1016/j.patcog.2025.112765"},"language":"en","primary_location":{"id":"doi:10.1016/j.patcog.2025.112765","doi":"https://doi.org/10.1016/j.patcog.2025.112765","is_oa":true,"landing_page_url":"https://doi.org/10.1016/j.patcog.2025.112765","pdf_url":null,"source":{"id":"https://openalex.org/S414566","display_name":"Pattern Recognition","issn_l":"0031-3203","issn":["0031-3203","1873-5142"],"is_oa":false,"is_in_doaj":false,"is_core":true,"listed_in":["anvur-scientific","ccf-b","cwts-core","dongbi-b","jufo-3","norway-2","poland-140","russia-white-list-1","vabb-shw"],"host_organization":"https://openalex.org/P4310320990","host_organization_name":"Elsevier BV","host_organization_lineage":["https://openalex.org/P4310320990"],"host_organization_lineage_names":["Elsevier BV"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Pattern Recognition","raw_type":"journal-article"},"volume":null,"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://doi.org/10.1016/j.patcog.2025.112765","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5114634767","display_name":"Nidhal Jegham","orcid":"https://orcid.org/0009-0001-1006-8127","observed_orcids":["https://orcid.org/0009-0001-1006-8127"]},"institutions":[{"id":"https://openalex.org/I108714496","display_name":"Tunis University","ror":"https://ror.org/02q1spa57","country_code":"TN","type":"education","lineage":["https://openalex.org/I108714496"]},{"id":"https://openalex.org/I17626003","display_name":"University of Rhode Island","ror":"https://ror.org/013ckk937","country_code":"US","type":"education","lineage":["https://openalex.org/I17626003"]}],"countries":["TN","US"],"is_corresponding":false,"raw_author_name":"Nidhal Jegham","raw_affiliation_strings":["Computer Science Department, University of Rhode Island, Kingston, RI, USA","Tunis Business School, University of Tunis, Tunis, Tunisia"],"raw_orcid":"https://orcid.org/0009-0001-1006-8127","affiliations":[{"raw_affiliation_string":"Computer Science Department, University of Rhode Island, Kingston, RI, USA","institution_ids":["https://openalex.org/I17626003"]},{"raw_affiliation_string":"Tunis Business School, University of Tunis, Tunis, Tunisia","institution_ids":["https://openalex.org/I108714496"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5076314302","display_name":"Marwan F. Abdelatti","orcid":"https://orcid.org/0000-0002-4782-6286","observed_orcids":["https://orcid.org/0000-0002-4782-6286"]},"institutions":[{"id":"https://openalex.org/I17626003","display_name":"University of Rhode Island","ror":"https://ror.org/013ckk937","country_code":"US","type":"education","lineage":["https://openalex.org/I17626003"]},{"id":"https://openalex.org/I196272386","display_name":"Providence College","ror":"https://ror.org/00rxpqe74","country_code":"US","type":"education","lineage":["https://openalex.org/I196272386"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Marwan Abdelatti","raw_affiliation_strings":["Computer Science Department, University of Rhode Island, Kingston, RI, USA","Department of Computer Science, Providence College, Providence, RI, USA"],"raw_orcid":"https://orcid.org/0000-0002-4782-6286","affiliations":[{"raw_affiliation_string":"Computer Science Department, University of Rhode Island, Kingston, RI, USA","institution_ids":["https://openalex.org/I17626003"]},{"raw_affiliation_string":"Department of Computer Science, Providence College, Providence, RI, USA","institution_ids":["https://openalex.org/I196272386"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5058249834","display_name":"Abdeltawab Hendawi","orcid":"https://orcid.org/0000-0003-3385-3379","observed_orcids":["https://orcid.org/0000-0003-3385-3379"]},"institutions":[{"id":"https://openalex.org/I17626003","display_name":"University of Rhode Island","ror":"https://ror.org/013ckk937","country_code":"US","type":"education","lineage":["https://openalex.org/I17626003"]}],"countries":["US"],"is_corresponding":true,"raw_author_name":"Abdeltawab Hendawi","raw_affiliation_strings":["Computer Science Department, University of Rhode Island, Kingston, RI, USA"],"raw_orcid":"https://orcid.org/0000-0003-3385-3379","affiliations":[{"raw_affiliation_string":"Computer Science Department, University of Rhode Island, Kingston, RI, USA","institution_ids":["https://openalex.org/I17626003"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":3,"corresponding_author_ids":["https://openalex.org/A5058249834"],"corresponding_institution_ids":["https://openalex.org/I17626003"],"apc_list":{"value":2590,"currency":"USD","value_usd":2590},"apc_paid":{"value":2590,"currency":"USD","value_usd":2590},"fwci":1.4954,"has_fulltext":false,"cited_by_count":4,"citation_normalized_percentile":{"value":0.83757728,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":98,"max":99},"biblio":{"volume":"172","issue":null,"first_page":"112765","last_page":"112765"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":1.0,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":1.0,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.0,"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/T10028","display_name":"Topic Modeling","score":0.0,"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/multimodal-large-language-models","display_name":"multimodal large language models","score":0.9829999804496765},{"id":"https://openalex.org/keywords/visual-reasoning","display_name":"visual reasoning","score":0.9819999933242798},{"id":"https://openalex.org/keywords/position-bias","display_name":"position bias","score":0.7879999876022339},{"id":"https://openalex.org/keywords/uncertainty-calibration","display_name":"uncertainty calibration","score":0.7639999985694885},{"id":"https://openalex.org/keywords/reasoning-consistency","display_name":"reasoning consistency","score":0.7149999737739563},{"id":"https://openalex.org/keywords/entropy-measures","display_name":"entropy measures","score":0.8130000233650208},{"id":"https://openalex.org/keywords/benchmark","display_name":"benchmark","score":0.49900001287460327},{"id":"https://openalex.org/keywords/gpt-4o","display_name":"GPT-4o","score":0.9350000023841858},{"id":"https://openalex.org/keywords/grok-3","display_name":"Grok 3","score":0.8009999990463257},{"id":"https://openalex.org/keywords/gemini-2-0-flash","display_name":"Gemini 2.0 Flash","score":0.7020000219345093}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6715999841690063},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.6281999945640564},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6236000061035156},{"id":"https://openalex.org/C106301342","wikidata":"https://www.wikidata.org/wiki/Q4117933","display_name":"Entropy (arrow of time)","level":2,"score":0.5622000098228455},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.5497000217437744},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.5394999980926514},{"id":"https://openalex.org/C2776436953","wikidata":"https://www.wikidata.org/wiki/Q5163215","display_name":"Consistency (knowledge bases)","level":2,"score":0.5250999927520752},{"id":"https://openalex.org/C176217482","wikidata":"https://www.wikidata.org/wiki/Q860554","display_name":"Metric (unit)","level":2,"score":0.5105000138282776},{"id":"https://openalex.org/C2777508537","wikidata":"https://www.wikidata.org/wiki/Q7936620","display_name":"Visual reasoning","level":2,"score":0.4925999939441681},{"id":"https://openalex.org/C2779290492","wikidata":"https://www.wikidata.org/wiki/Q6155940","display_name":"Janus","level":2,"score":0.33649998903274536},{"id":"https://openalex.org/C192209626","wikidata":"https://www.wikidata.org/wiki/Q190909","display_name":"Focus (optics)","level":2,"score":0.3149000108242035},{"id":"https://openalex.org/C20162079","wikidata":"https://www.wikidata.org/wiki/Q1151406","display_name":"Case-based reasoning","level":2,"score":0.2883000075817108}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1016/j.patcog.2025.112765","doi":"https://doi.org/10.1016/j.patcog.2025.112765","is_oa":true,"landing_page_url":"https://doi.org/10.1016/j.patcog.2025.112765","pdf_url":null,"source":{"id":"https://openalex.org/S414566","display_name":"Pattern Recognition","issn_l":"0031-3203","issn":["0031-3203","1873-5142"],"is_oa":false,"is_in_doaj":false,"is_core":true,"listed_in":["anvur-scientific","ccf-b","cwts-core","dongbi-b","jufo-3","norway-2","poland-140","russia-white-list-1","vabb-shw"],"host_organization":"https://openalex.org/P4310320990","host_organization_name":"Elsevier BV","host_organization_lineage":["https://openalex.org/P4310320990"],"host_organization_lineage_names":["Elsevier BV"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Pattern Recognition","raw_type":"journal-article"}],"best_oa_location":{"id":"doi:10.1016/j.patcog.2025.112765","doi":"https://doi.org/10.1016/j.patcog.2025.112765","is_oa":true,"landing_page_url":"https://doi.org/10.1016/j.patcog.2025.112765","pdf_url":null,"source":{"id":"https://openalex.org/S414566","display_name":"Pattern Recognition","issn_l":"0031-3203","issn":["0031-3203","1873-5142"],"is_oa":false,"is_in_doaj":false,"is_core":true,"listed_in":["anvur-scientific","ccf-b","cwts-core","dongbi-b","jufo-3","norway-2","poland-140","russia-white-list-1","vabb-shw"],"host_organization":"https://openalex.org/P4310320990","host_organization_name":"Elsevier BV","host_organization_lineage":["https://openalex.org/P4310320990"],"host_organization_lineage_names":["Elsevier BV"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Pattern Recognition","raw_type":"journal-article"},"sustainable_development_goals":[],"sustainable_development_goals_aurora":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":5,"referenced_works":["https://openalex.org/W4285255856","https://openalex.org/W4297253404","https://openalex.org/W4392570944","https://openalex.org/W4403997808","https://openalex.org/W4406757260"],"related_works":[],"abstract_inverted_index":{"Traditional":[0],"evaluations":[1],"of":[2,195],"multimodal":[3,186],"large":[4],"language":[5],"models":[6,130],"(LLMs)":[7],"have":[8],"been":[9],"limited":[10],"by":[11,35,119],"their":[12,160],"focus":[13],"on":[14],"single-image":[15],"reasoning,":[16],"failing":[17],"to":[18,72,77,162],"assess":[19,82],"crucial":[20],"aspects":[21],"like":[22],"contextual":[23],"understanding,":[24],"reasoning":[25,44,63,102],"stability,":[26],"and":[27,49,69,95,113,135,149,165,174,192],"uncertainty":[28,136],"calibration.":[29],"This":[30],"study":[31],"addresses":[32],"these":[33],"limitations":[34],"introducing":[36],"a":[37,58,74,181,189],"novel":[38,59],"benchmark":[39,179],"that":[40],"uniquely":[41],"integrates":[42],"multi-image":[43,170],"tasks":[45],"with":[46],"rejection-based":[47],"evaluation":[48],"positional":[50,163],"bias":[51,134,164],"detection.":[52],"We":[53,81],"further":[54],"introduce":[55],"entropy":[56,151],"as":[57,87],"metric":[60],"for":[61,184],"quantifying":[62],"consistency":[64,176],"across":[65,99],"reordered":[66],"answer":[67],"variants":[68],"abstention":[70],"rate":[71],"measure":[73],"model\u2019s":[75],"tendency":[76],"avoid":[78],"uncertain":[79],"answers.":[80],"8":[83],"state-of-the-art":[84],"models,":[85,98],"such":[86],"Grok":[88],"3,":[89],"ChatGPT-4o,":[90],"Gemini":[91,120],"2.0":[92,121],"Flash":[93,122],"Experimental,":[94],"DeepSeek\u2019s":[96],"Janus":[97,129,146,156],"eight":[100],"visual":[101],"tasks.":[103],"Our":[104],"findings":[105],"reveal":[106],"ChatGPT-o1":[107],"leading":[108],"in":[109,133,139],"overall":[110],"accuracy":[111,115],"(82.5%)":[112],"rejection":[114,141,172],"(70.0%),":[116],"closely":[117],"followed":[118],"Experimental":[123],"(70.8%).":[124],"On":[125],"the":[126],"other":[127],"hand,":[128],"exhibited":[131],"challenges":[132],"calibration,":[137],"reflected":[138],"low":[140],"accuracies":[142],"(Janus":[143,153],"7B:":[144,154],"25.0%,":[145],"1B:":[147,157],"30.0%)":[148],"high":[150],"scores":[152],"0.839,":[155],"0.787),":[158],"underscoring":[159],"susceptibility":[161],"unstable":[166],"reasoning.":[167],"By":[168],"employing":[169],"contexts,":[171],"mechanisms,":[173],"entropy-based":[175],"metrics,":[177],"this":[178],"sets":[180],"new":[182],"standard":[183],"evaluating":[185],"LLMs,":[187],"enabling":[188],"more":[190],"robust":[191],"reliable":[193],"assessment":[194],"next-generation":[196],"AI":[197],"systems.":[198]},"counts_by_year":[{"year":2026,"cited_by_count":4}],"updated_date":"2026-10-08T09:06:37.294493","created_date":"2025-11-20T00:00:00"}