{"id":"https://openalex.org/W7160110418","doi":"https://doi.org/10.48550/arxiv.2605.00591","title":"Intrinsic Gradient Suppression for Label-Noise Prompt Tuning in Vision-Language Models","display_name":"Intrinsic Gradient Suppression for Label-Noise Prompt Tuning in Vision-Language Models","publication_year":2026,"publication_date":"2026-05-01","ids":{"openalex":"https://openalex.org/W7160110418","doi":"https://doi.org/10.48550/arxiv.2605.00591"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.00591","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.00591","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.2605.00591","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5135101924","display_name":"Jiayu Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Jiayu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135207352","display_name":"Jiaxin Qi","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qi, Jiaxin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135145966","display_name":"Sheng Zhou","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhou, Sheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135224161","display_name":"Jiaqiang Huang","orcid":"https://orcid.org/0000-0001-8250-228X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Huang, Jiaqiang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5104579468","display_name":"Xiansheng Hua","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hua, Xiansheng","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":0.2084999978542328,"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":0.2084999978542328,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.19339999556541443,"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":0.164900004863739,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/robustness","display_name":"Robustness (evolution)","score":0.623199999332428},{"id":"https://openalex.org/keywords/probabilistic-logic","display_name":"Probabilistic logic","score":0.5317000150680542},{"id":"https://openalex.org/keywords/noise","display_name":"Noise (video)","score":0.3382999897003174},{"id":"https://openalex.org/keywords/gradient-descent","display_name":"Gradient descent","score":0.32600000500679016},{"id":"https://openalex.org/keywords/noisy-data","display_name":"Noisy data","score":0.29409998655319214},{"id":"https://openalex.org/keywords/saturation","display_name":"Saturation (graph theory)","score":0.28760001063346863}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6359000205993652},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.623199999332428},{"id":"https://openalex.org/C49937458","wikidata":"https://www.wikidata.org/wiki/Q2599292","display_name":"Probabilistic logic","level":2,"score":0.5317000150680542},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.44760000705718994},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4271000027656555},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.3382999897003174},{"id":"https://openalex.org/C153258448","wikidata":"https://www.wikidata.org/wiki/Q1199743","display_name":"Gradient descent","level":3,"score":0.32600000500679016},{"id":"https://openalex.org/C2781170535","wikidata":"https://www.wikidata.org/wiki/Q30587856","display_name":"Noisy data","level":2,"score":0.29409998655319214},{"id":"https://openalex.org/C9930424","wikidata":"https://www.wikidata.org/wiki/Q7426587","display_name":"Saturation (graph theory)","level":2,"score":0.28760001063346863},{"id":"https://openalex.org/C114289077","wikidata":"https://www.wikidata.org/wiki/Q3284399","display_name":"Statistical model","level":2,"score":0.27790001034736633},{"id":"https://openalex.org/C103088060","wikidata":"https://www.wikidata.org/wiki/Q1062839","display_name":"Error detection and correction","level":2,"score":0.2757999897003174},{"id":"https://openalex.org/C47446073","wikidata":"https://www.wikidata.org/wiki/Q5165890","display_name":"Control theory (sociology)","level":3,"score":0.27059999108314514},{"id":"https://openalex.org/C157524613","wikidata":"https://www.wikidata.org/wiki/Q2828883","display_name":"Fine-tuning","level":2,"score":0.2687999904155731},{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.25999999046325684}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.00591","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.00591","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.2605.00591","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.00591","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":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Contrastive":[0],"vision-language":[1],"models":[2],"like":[3],"CLIP":[4,34],"exhibit":[5],"remarkable":[6],"zero-shot":[7],"generalization.":[8],"However,":[9],"prompt":[10,126],"tuning":[11],"remains":[12],"highly":[13],"sensitive":[14],"to":[15],"label":[16],"noise,":[17],"as":[18],"mislabeled":[19],"samples":[20,89],"generate":[21],"disproportionately":[22],"large":[23],"gradients":[24,85],"that":[25,32,83,131],"can":[26],"overwhelm":[27],"pre-trained":[28],"priors.":[29],"We":[30,94],"argue":[31],"because":[33],"already":[35],"provides":[36],"a":[37,64,73,79,116,120],"near-optimal":[38],"initialization,":[39],"adaptation":[40],"should":[41],"be":[42],"inherently":[43],"conservative,":[44],"particularly":[45],"against":[46],"the":[47],"extreme":[48],"gradient":[49,69],"updates":[50],"common":[51],"in":[52],"noisy":[53,88,141],"settings.":[54],"To":[55],"this":[56,105,132],"end,":[57],"we":[58],"propose":[59],"Double-Softmax":[60],"Prompt":[61],"Tuning":[62],"(DSPT),":[63],"hyperparameter-free":[65],"method":[66],"for":[67,124],"intrinsic":[68],"suppression.":[70,109],"By":[71],"applying":[72],"sequential":[74],"probabilistic":[75],"normalization,":[76],"DSPT":[77],"induces":[78],"self-adaptive":[80],"saturation":[81],"zone":[82],"suppresses":[84],"from":[86],"high-error":[87],"while":[90],"maintaining":[91],"informative":[92],"updates.":[93],"also":[95],"provide":[96],"both":[97],"theoretical":[98],"analysis":[99],"and":[100,148],"empirical":[101],"evidence":[102],"about":[103],"how":[104],"mechanism":[106],"achieves":[107,136],"adaptive":[108],"This":[110],"design":[111,135],"transforms":[112],"``gradient":[113],"vanishing'',":[114],"traditionally":[115],"training":[117],"bottleneck,":[118],"into":[119],"principled":[121],"noise-filtering":[122],"shield":[123],"label-noise":[125],"tuning.":[127],"Extensive":[128],"experiments":[129],"confirm":[130],"simple,":[133],"drop-in":[134],"state-of-the-art":[137],"robustness":[138],"across":[139],"various":[140],"benchmarks,":[142],"outperforming":[143],"methods":[144],"with":[145],"complex":[146],"architectures":[147],"handcrafted":[149],"hyperparameters.":[150]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-05T00:00:00"}