{"id":"https://openalex.org/W7161730233","doi":"https://doi.org/10.48550/arxiv.2605.17144","title":"Contrastive Conceptor Activation Steering (COAST): Unlocking Vision-Language-Action Models through Hidden States","display_name":"Contrastive Conceptor Activation Steering (COAST): Unlocking Vision-Language-Action Models through Hidden States","publication_year":2026,"publication_date":"2026-05-16","ids":{"openalex":"https://openalex.org/W7161730233","doi":"https://doi.org/10.48550/arxiv.2605.17144"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.17144","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.17144","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":"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.2605.17144","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5119181466","display_name":"Miranda Muqing Miao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Miao, Miranda Muqing","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136478511","display_name":"Subin Kim","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kim, Subin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5084812990","display_name":"Brandon Yang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Brandon","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5136471504","display_name":"Lyle Ungar","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ungar, Lyle","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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.5917999744415283,"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":0.5917999744415283,"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/T10709","display_name":"Social Robot Interaction and HRI","score":0.10270000249147415,"subfield":{"id":"https://openalex.org/subfields/3207","display_name":"Social Psychology"},"field":{"id":"https://openalex.org/fields/32","display_name":"Psychology"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},{"id":"https://openalex.org/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.07750000059604645,"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/leverage","display_name":"Leverage (statistics)","score":0.6011000275611877},{"id":"https://openalex.org/keywords/bottleneck","display_name":"Bottleneck","score":0.4562999904155731},{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.43869999051094055},{"id":"https://openalex.org/keywords/cohesion","display_name":"Cohesion (chemistry)","score":0.42719998955726624},{"id":"https://openalex.org/keywords/perception","display_name":"Perception","score":0.3831999897956848},{"id":"https://openalex.org/keywords/subspace-topology","display_name":"Subspace topology","score":0.36640000343322754},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.350600004196167},{"id":"https://openalex.org/keywords/linear-subspace","display_name":"Linear subspace","score":0.3483999967575073}],"concepts":[{"id":"https://openalex.org/C153083717","wikidata":"https://www.wikidata.org/wiki/Q6535263","display_name":"Leverage (statistics)","level":2,"score":0.6011000275611877},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5450999736785889},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5008999705314636},{"id":"https://openalex.org/C2780513914","wikidata":"https://www.wikidata.org/wiki/Q18210350","display_name":"Bottleneck","level":2,"score":0.4562999904155731},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.43869999051094055},{"id":"https://openalex.org/C104054115","wikidata":"https://www.wikidata.org/wiki/Q216828","display_name":"Cohesion (chemistry)","level":2,"score":0.42719998955726624},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3874000012874603},{"id":"https://openalex.org/C26760741","wikidata":"https://www.wikidata.org/wiki/Q160402","display_name":"Perception","level":2,"score":0.3831999897956848},{"id":"https://openalex.org/C32834561","wikidata":"https://www.wikidata.org/wiki/Q660730","display_name":"Subspace topology","level":2,"score":0.36640000343322754},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.350600004196167},{"id":"https://openalex.org/C12362212","wikidata":"https://www.wikidata.org/wiki/Q728435","display_name":"Linear subspace","level":2,"score":0.3483999967575073},{"id":"https://openalex.org/C159877910","wikidata":"https://www.wikidata.org/wiki/Q2202883","display_name":"Autoregressive model","level":2,"score":0.33329999446868896},{"id":"https://openalex.org/C2777735758","wikidata":"https://www.wikidata.org/wiki/Q817765","display_name":"Path (computing)","level":2,"score":0.319599986076355},{"id":"https://openalex.org/C155512373","wikidata":"https://www.wikidata.org/wiki/Q287450","display_name":"Residual","level":2,"score":0.31310001015663147},{"id":"https://openalex.org/C114289077","wikidata":"https://www.wikidata.org/wiki/Q3284399","display_name":"Statistical model","level":2,"score":0.31060001254081726},{"id":"https://openalex.org/C17020691","wikidata":"https://www.wikidata.org/wiki/Q139677","display_name":"Operator (biology)","level":5,"score":0.3102000057697296},{"id":"https://openalex.org/C161840515","wikidata":"https://www.wikidata.org/wiki/Q186131","display_name":"Terrain","level":2,"score":0.2935999929904938},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.2757999897003174},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.27250000834465027},{"id":"https://openalex.org/C177769412","wikidata":"https://www.wikidata.org/wiki/Q278090","display_name":"Prior probability","level":3,"score":0.25999999046325684}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.17144","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.17144","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":"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.2605.17144","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.17144","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":"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":[{"score":0.8288728594779968,"display_name":"Life below water","id":"https://metadata.un.org/sdg/14"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Vision-Language-Action":[0],"(VLA)":[1],"models":[2],"leverage":[3],"powerful":[4],"perceptual":[5],"priors":[6],"from":[7,70],"web-scale":[8],"Vision-Language":[9],"Model":[10],"(VLM)":[11],"pre-training,":[12],"yet":[13],"they":[14],"remain":[15,140],"surprisingly":[16],"brittle":[17],"in":[18,174],"practice,":[19],"frequently":[20],"failing":[21],"at":[22],"simple":[23],"robotic":[24,68],"tasks.":[25],"To":[26],"mitigate":[27],"this,":[28],"we":[29],"propose":[30],"Contrastive":[31],"Conceptor":[32],"Activation":[33],"Steering":[34],"(COAST).":[35],"COAST":[36,58,108,196],"builds":[37],"on":[38,158],"the":[39,51,180,209],"notion":[40],"of":[41,54,74],"a":[42,44,55,66,71,198],"\"conceptor\",":[43],"linear":[45],"operator":[46],"that":[47,129,167,179],"soft-projects":[48],"data":[49],"into":[50,86],"principal":[52],"components":[53],"target":[56,67],"distribution.":[57],"uses":[59],"conceptors":[60,154],"to":[61,91,155,202],"identify":[62],"success-critical":[63],"subspaces":[64,90],"for":[65],"task":[69,93,115],"few":[72],"examples":[73],"success":[75,89,116,138],"and":[76,105,113,121,178],"failure":[77,130,147],"rollouts.":[78],"At":[79],"inference":[80],"time,":[81],"it":[82],"steers":[83],"VLA":[84],"latents":[85],"these":[87,204],"identified":[88],"improve":[92,156],"outcomes.":[94],"Across":[95],"three":[96],"architecturally":[97],"distinct":[98],"neural":[99],"policies":[100],"(flow-matching":[101],"VLA,":[102,104],"autoregressive":[103],"Diffusion":[106],"Policy),":[107],"improves":[109],"absolute":[110],"mean":[111],"simulation":[112],"real-robot":[114],"rate":[117],"by":[118,188,207],"over":[119],"20":[120],"40%":[122],"respectively.":[123],"The":[124],"activation":[125],"subspace":[126],"geometry":[127],"reveals":[128],"modes":[131],"share":[132,145],"substantial":[133,171],"structure":[134,150],"across":[135],"tasks":[136,144,160],"while":[137],"representations":[139],"largely":[141],"task-specific.":[142],"When":[143],"similar":[146],"modes,":[148],"this":[149],"enables":[151],"previously":[152],"fitted":[153],"performance":[157],"new":[159],"without":[161],"refitting.":[162],"Ultimately,":[163],"our":[164],"results":[165],"suggest":[166],"current":[168],"VLAs":[169],"retain":[170],"task-relevant":[172,194],"knowledge":[173],"their":[175],"latent":[176,205],"representations,":[177],"action":[181],"expert's":[182],"decoding":[183],"bottleneck":[184],"could":[185],"be":[186],"mitigated":[187],"steering":[189,208],"its":[190,212],"residual":[191],"stream":[192],"toward":[193],"subspaces.":[195],"provides":[197],"lightweight,":[199],"training-free":[200],"path":[201],"unlocking":[203],"capabilities":[206],"model":[210],"towards":[211],"own":[213],"\"success\"":[214],"distributions.":[215]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-05-20T00:00:00"}