{"id":"https://openalex.org/W4416613142","doi":"https://doi.org/10.48550/arxiv.2505.23723","title":"ML-Agent: Reinforcing LLM Agents for Autonomous Machine Learning Engineering","display_name":"ML-Agent: Reinforcing LLM Agents for Autonomous Machine Learning Engineering","publication_year":2025,"publication_date":"2025-05-29","ids":{"openalex":"https://openalex.org/W4416613142","doi":"https://doi.org/10.48550/arxiv.2505.23723"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2505.23723","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2505.23723","pdf_url":"https://arxiv.org/pdf/2505.23723","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,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2505.23723","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5026487999","display_name":"Zihang Liu","orcid":"https://orcid.org/0009-0000-9876-5244","observed_orcids":["https://orcid.org/0009-0000-9876-5244"]},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]},{"id":"https://openalex.org/I4391012619","display_name":"Shanghai Artificial Intelligence Laboratory","ror":"https://ror.org/03wkvpx79","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4391012619"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Liu, Zexi","raw_affiliation_strings":["Shanghai AI Laboratory","Shanghai Jiao Tong University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shanghai AI Laboratory","institution_ids":["https://openalex.org/I4391012619"]},{"raw_affiliation_string":"Shanghai Jiao Tong University","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5064810450","display_name":"Jingyi Chai","orcid":"https://orcid.org/0009-0000-2959-3274","observed_orcids":["https://orcid.org/0009-0000-2959-3274"]},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]},{"id":"https://openalex.org/I4391012619","display_name":"Shanghai Artificial Intelligence Laboratory","ror":"https://ror.org/03wkvpx79","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4391012619"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Chai, Jingyi","raw_affiliation_strings":["Shanghai AI Laboratory","Shanghai Jiao Tong University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shanghai AI Laboratory","institution_ids":["https://openalex.org/I4391012619"]},{"raw_affiliation_string":"Shanghai Jiao Tong University","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101532554","display_name":"Xinyu Zhu","orcid":"https://orcid.org/0000-0002-5352-9499","observed_orcids":["https://orcid.org/0000-0002-5352-9499","https://orcid.org/0009-0000-0341-9032"]},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhu, Xinyu","raw_affiliation_strings":["Shanghai Jiao Tong University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5016880107","display_name":"Shuo Tang","orcid":"https://orcid.org/0000-0002-7124-9167","observed_orcids":["https://orcid.org/0000-0002-7124-9167"]},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Tang, Shuo","raw_affiliation_strings":["Shanghai Jiao Tong University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100728766","display_name":"Rui Ye","orcid":"https://orcid.org/0009-0007-5998-8200","observed_orcids":["https://orcid.org/0009-0007-5998-8200"]},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Ye, Rui","raw_affiliation_strings":["Shanghai Jiao Tong University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100770109","display_name":"Bo Zhang","orcid":"https://orcid.org/0000-0001-8052-782X","observed_orcids":["https://orcid.org/0000-0001-8052-782X","https://orcid.org/0000-0002-9795-4673","https://orcid.org/0009-0002-0171-4905","https://orcid.org/0009-0005-3354-8018"]},"institutions":[{"id":"https://openalex.org/I4391012619","display_name":"Shanghai Artificial Intelligence Laboratory","ror":"https://ror.org/03wkvpx79","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4391012619"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhang, Bo","raw_affiliation_strings":["Shanghai AI Laboratory"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shanghai AI Laboratory","institution_ids":["https://openalex.org/I4391012619"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5008619721","display_name":"Lei Bai","orcid":"https://orcid.org/0000-0001-9110-5737","observed_orcids":["https://orcid.org/0000-0001-9110-5737"]},"institutions":[{"id":"https://openalex.org/I4391012619","display_name":"Shanghai Artificial Intelligence Laboratory","ror":"https://ror.org/03wkvpx79","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4391012619"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Bai, Lei","raw_affiliation_strings":["Shanghai AI Laboratory"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shanghai AI Laboratory","institution_ids":["https://openalex.org/I4391012619"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5066373402","display_name":"Siheng Chen","orcid":"https://orcid.org/0000-0002-9144-0583","observed_orcids":["https://orcid.org/0000-0002-9144-0583"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Siheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.5976999998092651,"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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.5976999998092651,"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/T13904","display_name":"Artificial Intelligence Applications","score":0.14669999480247498,"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/T12072","display_name":"Machine Learning and Algorithms","score":0.03519999980926514,"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/llm-based-agents","display_name":"LLM agents","score":0.8560000061988831},{"id":"https://openalex.org/keywords/autonomous-machine-learning","display_name":"autonomous machine learning","score":0.9240000247955322},{"id":"https://openalex.org/keywords/online-reinforcement-learning","display_name":"online reinforcement learning","score":0.847000002861023},{"id":"https://openalex.org/keywords/reward-function-design","display_name":"reward function design","score":0.6349999904632568},{"id":"https://openalex.org/keywords/qwen2-5","display_name":"Qwen2.5","score":0.7839999794960022},{"id":"https://openalex.org/keywords/cross-task-generalization","display_name":"cross-task generalization","score":0.796999990940094}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7340999841690063},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.7315000295639038},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6636000275611877},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.5950999855995178},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.5382999777793884},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.5164999961853027},{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.49810001254081726},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.38280001282691956},{"id":"https://openalex.org/C13687954","wikidata":"https://www.wikidata.org/wiki/Q4826847","display_name":"Autonomous agent","level":2,"score":0.376800000667572},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.3379000127315521},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.3370000123977661},{"id":"https://openalex.org/C77967617","wikidata":"https://www.wikidata.org/wiki/Q4677561","display_name":"Active learning (machine learning)","level":2,"score":0.32109999656677246},{"id":"https://openalex.org/C34413123","wikidata":"https://www.wikidata.org/wiki/Q170978","display_name":"Robotics","level":3,"score":0.2962000072002411},{"id":"https://openalex.org/C47932503","wikidata":"https://www.wikidata.org/wiki/Q5395689","display_name":"Error-driven learning","level":3,"score":0.28360000252723694},{"id":"https://openalex.org/C2986563244","wikidata":"https://www.wikidata.org/wiki/Q6822310","display_name":"Learning to learn","level":2,"score":0.2565999925136566}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2505.23723","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2505.23723","pdf_url":"https://arxiv.org/pdf/2505.23723","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,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2505.23723","doi":"https://doi.org/10.48550/arxiv.2505.23723","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2505.23723","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,"listed_in":[],"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":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2505.23723","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2505.23723","pdf_url":"https://arxiv.org/pdf/2505.23723","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,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4416613142.pdf","grobid_xml":"https://content.openalex.org/works/W4416613142.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4416023759","https://openalex.org/W7222632291","https://openalex.org/W4417517367","https://openalex.org/W7128543922","https://openalex.org/W4417536788","https://openalex.org/W4416169577","https://openalex.org/W4411464987","https://openalex.org/W4387432648","https://openalex.org/W4416385763","https://openalex.org/W4403885480"],"abstract_inverted_index":{"The":[0],"emergence":[1],"of":[2,13,61],"large":[3,39],"language":[4],"model":[5],"(LLM)-based":[6],"agents":[7,102,176],"has":[8],"significantly":[9,186],"advanced":[10],"the":[11,20,29,54,59],"development":[12],"autonomous":[14,160],"machine":[15],"learning":[16,79],"(ML)":[17],"engineering.":[18],"However,":[19],"dominant":[21],"prompt-based":[22],"paradigm":[23,60],"exhibits":[24],"limitations:":[25],"smaller":[26],"models":[27,41],"lack":[28],"capacity":[30],"to":[31,103,175],"learn":[32],"from":[33],"execution":[34],"trajectories":[35],"for":[36,53,107,144,159],"generalization,":[37],"while":[38],"proprietary":[40,180],"incur":[42],"high":[43],"computational":[44,188],"overhead,":[45],"restricting":[46],"accessibility":[47],"and":[48,125,193],"scalability.":[49],"Focusing":[50],"on":[51,73,117,164],"this,":[52,83],"first":[55],"time,":[56],"we":[57,84,150],"explore":[58],"learning-based":[62],"agentic":[63,88,131],"ML,":[64],"where":[65],"an":[66,130],"LLM":[67,101,158],"agent":[68],"learns":[69],"through":[70],"interactive":[71],"experimentation":[72],"ML":[74,89,138,167],"tasks":[75],"using":[76,177],"online":[77],"reinforcement":[78],"(RL).":[80],"To":[81],"realize":[82],"propose":[85],"a":[86,118,155],"novel":[87],"training":[90,116,127,163],"framework":[91],"with":[92],"three":[93],"key":[94],"components:":[95],"(1)":[96],"exploration-enriched":[97],"fine-tuning,":[98],"which":[99,114,135],"enables":[100,115],"generate":[104],"diverse":[105],"actions":[106],"enhanced":[108],"RL":[109,145],"exploration;":[110],"(2)":[111],"step-wise":[112],"RL,":[113],"single":[119],"action":[120],"step,":[121],"accelerating":[122],"experience":[123],"collection":[124],"improving":[126],"efficiency;":[128],"(3)":[129],"ML-specific":[132],"reward":[133],"module,":[134],"unifies":[136],"varied":[137],"feedback":[139],"signals":[140],"into":[141],"consistent":[142],"rewards":[143],"optimization.":[146],"Leveraging":[147],"this":[148],"framework,":[149],"train":[151],"ML-Agent,":[152],"driven":[153],"by":[154],"7B-sized":[156,170],"Qwen-2.5":[157],"ML.":[161],"Despite":[162],"only":[165],"9":[166],"tasks,":[168],"our":[169],"ML-Agent":[171],"achieves":[172],"comparable":[173],"performance":[174,192],"much":[178],"larger":[179],"LLMs":[181],"(e.g.,":[182],"GPT-5)":[183],"but":[184],"at":[185],"lower":[187],"cost,":[189],"demonstrating":[190],"strong":[191],"cross-task":[194],"generalization.":[195]},"counts_by_year":[],"updated_date":"2026-10-30T12:39:25.321706","created_date":"2025-10-10T00:00:00"}