{"id":"https://openalex.org/W7134841662","doi":"https://doi.org/10.48550/arxiv.2603.07135","title":"The Model Knows Which Tokens Matter: Automatic Token Selection via Noise Gating","display_name":"The Model Knows Which Tokens Matter: Automatic Token Selection via Noise Gating","publication_year":2026,"publication_date":"2026-03-07","ids":{"openalex":"https://openalex.org/W7134841662","doi":"https://doi.org/10.48550/arxiv.2603.07135"},"language":null,"primary_location":{"id":"pmh:doi:10.48550/arxiv.2603.07135","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},"type":"article","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":null,"any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5128646065","display_name":"Landi He","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"He, Landi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128648376","display_name":"Xiaoyu Yang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Xiaoyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5128636684","display_name":"Lijian Xu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xu, Lijian","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":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.3077199,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"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":0.9562000036239624,"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":0.9562000036239624,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.007499999832361937,"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/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.002899999963119626,"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/security-token","display_name":"Security token","score":0.6464999914169312},{"id":"https://openalex.org/keywords/pruning","display_name":"Pruning","score":0.5999000072479248},{"id":"https://openalex.org/keywords/noise","display_name":"Noise (video)","score":0.5443000197410583},{"id":"https://openalex.org/keywords/selection","display_name":"Selection (genetic algorithm)","score":0.5378999710083008},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.5271999835968018},{"id":"https://openalex.org/keywords/diagonal","display_name":"Diagonal","score":0.4896000027656555},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.47600001096725464},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.3953999876976013},{"id":"https://openalex.org/keywords/variance","display_name":"Variance (accounting)","score":0.39309999346733093}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8169999718666077},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.6464999914169312},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.5999000072479248},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.5443000197410583},{"id":"https://openalex.org/C81917197","wikidata":"https://www.wikidata.org/wiki/Q628760","display_name":"Selection (genetic algorithm)","level":2,"score":0.5378999710083008},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5271999835968018},{"id":"https://openalex.org/C130367717","wikidata":"https://www.wikidata.org/wiki/Q189791","display_name":"Diagonal","level":2,"score":0.4896000027656555},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.4844000041484833},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.47600001096725464},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.3953999876976013},{"id":"https://openalex.org/C196083921","wikidata":"https://www.wikidata.org/wiki/Q7915758","display_name":"Variance (accounting)","level":2,"score":0.39309999346733093},{"id":"https://openalex.org/C93959086","wikidata":"https://www.wikidata.org/wiki/Q6888345","display_name":"Model selection","level":2,"score":0.36640000343322754},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.3474999964237213},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.3449999988079071},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.3434000015258789},{"id":"https://openalex.org/C2776257435","wikidata":"https://www.wikidata.org/wiki/Q1576430","display_name":"Bandwidth (computing)","level":2,"score":0.33559998869895935},{"id":"https://openalex.org/C127162648","wikidata":"https://www.wikidata.org/wiki/Q16858953","display_name":"Channel (broadcasting)","level":2,"score":0.335099995136261},{"id":"https://openalex.org/C190839683","wikidata":"https://www.wikidata.org/wiki/Q2448197","display_name":"Train","level":2,"score":0.33500000834465027},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3312000036239624},{"id":"https://openalex.org/C103278499","wikidata":"https://www.wikidata.org/wiki/Q254465","display_name":"Similarity (geometry)","level":3,"score":0.3208000063896179},{"id":"https://openalex.org/C2775973920","wikidata":"https://www.wikidata.org/wiki/Q3252726","display_name":"Selection algorithm","level":3,"score":0.3066999912261963},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.3027999997138977},{"id":"https://openalex.org/C43126263","wikidata":"https://www.wikidata.org/wiki/Q128751","display_name":"Source code","level":2,"score":0.29440000653266907},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.28839999437332153},{"id":"https://openalex.org/C45235069","wikidata":"https://www.wikidata.org/wiki/Q278425","display_name":"Table (database)","level":2,"score":0.28679999709129333},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2680000066757202},{"id":"https://openalex.org/C114289077","wikidata":"https://www.wikidata.org/wiki/Q3284399","display_name":"Statistical model","level":2,"score":0.2660999894142151},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.265500009059906},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.2612000107765198},{"id":"https://openalex.org/C29265498","wikidata":"https://www.wikidata.org/wiki/Q7047719","display_name":"Noise measurement","level":3,"score":0.25929999351501465},{"id":"https://openalex.org/C155846161","wikidata":"https://www.wikidata.org/wiki/Q1143367","display_name":"Graphical model","level":2,"score":0.2565999925136566},{"id":"https://openalex.org/C126042441","wikidata":"https://www.wikidata.org/wiki/Q1324888","display_name":"Frame (networking)","level":2,"score":0.25110000371932983}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:doi:10.48550/arxiv.2603.07135","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},{"id":"doi:10.48550/arxiv.2603.07135","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.07135","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:doi:10.48550/arxiv.2603.07135","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","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":{"Visual":[0],"tokens":[1],"dominate":[2],"inference":[3],"cost":[4],"in":[5],"vision-language":[6],"models":[7],"(VLMs),":[8],"yet":[9],"many":[10],"carry":[11],"redundant":[12],"information.":[13,52],"Existing":[14],"pruning":[15,32],"methods":[16],"alleviate":[17],"this":[18],"but":[19],"typically":[20],"rely":[21],"on":[22],"attention":[23,109],"magnitude":[24],"or":[25,80],"similarity":[26],"scores.":[27],"We":[28,53],"reformulate":[29],"visual":[30,51],"token":[31,74],"as":[33],"capacity":[34],"constrained":[35],"communication:":[36],"given":[37],"a":[38,58,64,85,107,122],"fixed":[39],"budget":[40],"K,":[41],"the":[42,71,113,119],"model":[43,139],"must":[44],"allocate":[45],"limited":[46],"bandwidth":[47],"to":[48,63,96,154],"maximally":[49],"preserve":[50],"propose":[54],"AutoSelect,":[55],"which":[56],"attaches":[57],"lightweight":[59],"Scorer":[60,120],"and":[61,67,121,152],"Denoiser":[62,110],"frozen":[65],"VLM":[66,132,156],"trains":[68],"with":[69,147],"only":[70,118,148],"standard":[72],"next":[73],"prediction":[75],"loss,":[76],"without":[77,158],"auxiliary":[78],"objectives":[79],"extra":[81],"annotations.":[82],"During":[83],"training,":[84],"variance":[86],"preserving":[87],"noise":[88],"gate":[89],"modulates":[90],"each":[91],"token's":[92],"information":[93],"flow":[94],"according":[95],"its":[97],"predicted":[98],"importance":[99],"so":[100],"that":[101],"gradients":[102],"propagate":[103],"through":[104],"all":[105],"tokens;":[106],"diagonal":[108],"then":[111],"recovers":[112],"perturbed":[114],"representations.":[115],"At":[116],"inference,":[117],"hard":[123],"top-K":[124],"selection":[125],"remain,":[126],"adding":[127],"negligible":[128],"latency.":[129],"On":[130],"ten":[131],"benchmarks,":[133],"AutoSelect":[134],"retains":[135],"96.5%":[136],"of":[137],"full":[138],"accuracy":[140],"while":[141],"accelerating":[142],"LLM":[143],"prefill":[144],"by":[145],"2.85x":[146],"0.69":[149],"ms":[150],"overhead,":[151],"transfers":[153],"different":[155],"backbones":[157],"architecture-specific":[159],"tuning.":[160],"Code":[161],"is":[162],"available":[163],"at":[164],"https://github.com/MedHK23/AutoSelect.":[165]},"counts_by_year":[],"updated_date":"2026-07-15T18:14:33.161393","created_date":"2026-03-11T00:00:00"}