{"id":"https://openalex.org/W7162466152","doi":"https://doi.org/10.48550/arxiv.2605.25143","title":"Beyond the Frontier: Stochastic Backtracking for Efficient Test-Time Scaling","display_name":"Beyond the Frontier: Stochastic Backtracking for Efficient Test-Time Scaling","publication_year":2026,"publication_date":"2026-05-24","ids":{"openalex":"https://openalex.org/W7162466152","doi":"https://doi.org/10.48550/arxiv.2605.25143"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.25143","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.25143","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.2605.25143","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5137037097","display_name":"Dao Tran","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tran, Dao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137053929","display_name":"Duc Anh Le","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Le, Duc Anh","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137008741","display_name":"Ngoc Luu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Luu, Ngoc","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136996542","display_name":"Quan Pham","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Pham, Quan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137003047","display_name":"Tung Pham","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Pham, Tung","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5137068714","display_name":"Hung Bui","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bui, Hung","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/T10028","display_name":"Topic Modeling","score":0.37560001015663147,"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/T10028","display_name":"Topic Modeling","score":0.37560001015663147,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.23330000042915344,"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/T13629","display_name":"Text Readability and Simplification","score":0.05590000003576279,"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/backtracking","display_name":"Backtracking","score":0.7160999774932861},{"id":"https://openalex.org/keywords/prefix","display_name":"Prefix","score":0.6198999881744385},{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.5993000268936157},{"id":"https://openalex.org/keywords/look-ahead","display_name":"Look-ahead","score":0.5358999967575073},{"id":"https://openalex.org/keywords/fraction","display_name":"Fraction (chemistry)","score":0.4869999885559082},{"id":"https://openalex.org/keywords/selection","display_name":"Selection (genetic algorithm)","score":0.4763999879360199},{"id":"https://openalex.org/keywords/reduction","display_name":"Reduction (mathematics)","score":0.41679999232292175},{"id":"https://openalex.org/keywords/scaling","display_name":"Scaling","score":0.39259999990463257},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.3862999975681305}],"concepts":[{"id":"https://openalex.org/C156884757","wikidata":"https://www.wikidata.org/wiki/Q798554","display_name":"Backtracking","level":2,"score":0.7160999774932861},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6462000012397766},{"id":"https://openalex.org/C141603448","wikidata":"https://www.wikidata.org/wiki/Q134830","display_name":"Prefix","level":2,"score":0.6198999881744385},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.5993000268936157},{"id":"https://openalex.org/C147297375","wikidata":"https://www.wikidata.org/wiki/Q6674930","display_name":"Look-ahead","level":2,"score":0.5358999967575073},{"id":"https://openalex.org/C149629883","wikidata":"https://www.wikidata.org/wiki/Q660926","display_name":"Fraction (chemistry)","level":2,"score":0.4869999885559082},{"id":"https://openalex.org/C81917197","wikidata":"https://www.wikidata.org/wiki/Q628760","display_name":"Selection (genetic algorithm)","level":2,"score":0.4763999879360199},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.4512999951839447},{"id":"https://openalex.org/C111335779","wikidata":"https://www.wikidata.org/wiki/Q3454686","display_name":"Reduction (mathematics)","level":2,"score":0.41679999232292175},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.4059000015258789},{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.39259999990463257},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.3862999975681305},{"id":"https://openalex.org/C19499675","wikidata":"https://www.wikidata.org/wiki/Q232207","display_name":"Monte Carlo method","level":2,"score":0.3677999973297119},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.34220001101493835},{"id":"https://openalex.org/C150921843","wikidata":"https://www.wikidata.org/wiki/Q1170431","display_name":"Resampling","level":2,"score":0.3379000127315521},{"id":"https://openalex.org/C120524526","wikidata":"https://www.wikidata.org/wiki/Q1709148","display_name":"Reboot","level":2,"score":0.33000001311302185},{"id":"https://openalex.org/C8272713","wikidata":"https://www.wikidata.org/wiki/Q176737","display_name":"Stochastic process","level":2,"score":0.2930000126361847},{"id":"https://openalex.org/C52740198","wikidata":"https://www.wikidata.org/wiki/Q1539564","display_name":"Importance sampling","level":3,"score":0.2921999990940094},{"id":"https://openalex.org/C194387892","wikidata":"https://www.wikidata.org/wiki/Q1747770","display_name":"Stochastic optimization","level":2,"score":0.2824999988079071},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.2800999879837036},{"id":"https://openalex.org/C2780586882","wikidata":"https://www.wikidata.org/wiki/Q7520643","display_name":"Simple (philosophy)","level":2,"score":0.2766000032424927},{"id":"https://openalex.org/C61797465","wikidata":"https://www.wikidata.org/wiki/Q1188986","display_name":"Term (time)","level":2,"score":0.26600000262260437},{"id":"https://openalex.org/C127491075","wikidata":"https://www.wikidata.org/wiki/Q7617825","display_name":"Stochastic modelling","level":2,"score":0.26570001244544983},{"id":"https://openalex.org/C118671147","wikidata":"https://www.wikidata.org/wiki/Q578714","display_name":"Quantile","level":2,"score":0.25600001215934753},{"id":"https://openalex.org/C163258240","wikidata":"https://www.wikidata.org/wiki/Q25342","display_name":"Power (physics)","level":2,"score":0.25519999861717224},{"id":"https://openalex.org/C148220186","wikidata":"https://www.wikidata.org/wiki/Q7111912","display_name":"Outcome (game theory)","level":2,"score":0.2547999918460846},{"id":"https://openalex.org/C187691185","wikidata":"https://www.wikidata.org/wiki/Q2020720","display_name":"Grid","level":2,"score":0.2531000077724457},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.25049999356269836}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.25143","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.25143","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.2605.25143","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.25143","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":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Test-time":[0],"scaling":[1],"improves":[2],"language":[3],"model":[4,164],"reasoning":[5,161],"by":[6,123],"spending":[7],"additional":[8],"compute":[9,95],"to":[10,19,38,96,135,148,191,206],"explore":[11],"multiple":[12],"solution":[13],"trajectories.":[14],"The":[15],"key":[16],"challenge":[17],"is":[18],"maximize":[20],"accuracy":[21,171,180],"while":[22],"minimizing":[23],"the":[24,49,59,73,105,149,176,186,208],"total":[25],"number":[26],"of":[27,75,90,102,179,185],"generated":[28,99],"tokens":[29],"during":[30],"reasoning.":[31],"Recent":[32],"PRM-guided":[33,121,193],"methods":[34,167],"score":[35],"intermediate":[36],"prefixes":[37,52,76,132],"steer":[39],"this":[40,110],"search,":[41],"but":[42],"most":[43],"are":[44],"frontier-only:":[45],"they":[46],"keep":[47],"only":[48,103,182],"current":[50,106],"active":[51],"and":[53,72,156,163,175,203],"irreversibly":[54],"prune":[55],"or":[56],"resample":[57],"away":[58],"rest":[60],"using":[61,152,181],"noisy":[62],"PRM":[63,154],"scores.":[64],"This":[65],"can":[66],"cause":[67],"premature":[68],"commitment,":[69],"diversity":[70],"collapse,":[71],"loss":[74],"that":[77,196],"still":[78],"admit":[79],"correct":[80],"continuations.":[81],"We":[82],"introduce":[83],"stochastic":[84,198],"backtracking":[85,199],"over":[86],"a":[87,133,183,201],"persistent":[88,150],"pool":[89,151],"historical":[91,131],"prefixes,":[92],"allowing":[93],"test-time":[94,212],"revisit":[97],"previously":[98],"states":[100],"instead":[101],"expanding":[104],"frontier.":[107],"To":[108],"make":[109],"efficient,":[111],"we":[112],"propose":[113],"two":[114],"complementary":[115],"mechanisms.":[116],"Subpool":[117],"Selection":[118],"strengthens":[119],"greedy":[120],"search":[122],"applying":[124],"Top-N":[125],"selection":[126],"within":[127],"random":[128],"subpools,":[129],"giving":[130],"chance":[134],"bypass":[136],"over-scored":[137],"frontier":[138],"candidates.":[139],"Power":[140],"Backtrack":[141],"Sequential":[142],"Monte":[143],"Carlo":[144],"extends":[145],"SMC-style":[146],"resampling":[147],"powered":[153],"scores":[155],"mixture-corrected":[157],"weights.":[158],"Across":[159],"mathematical":[160],"benchmarks":[162],"scales,":[165],"our":[166],"consistently":[168],"achieve":[169],"higher":[170],"per":[172],"token":[173,187],"count,":[174],"same":[177],"level":[178],"fraction":[184],"count":[188],"in":[189,211],"comparison":[190],"strong":[192],"baselines,":[194],"demonstrating":[195],"persistent-pool":[197],"provides":[200],"simple":[202],"effective":[204],"way":[205],"improve":[207],"accuracy-token":[209],"trade-off":[210],"scaling.":[213]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-05-27T00:00:00"}