{"id":"https://openalex.org/W3023426586","doi":"https://doi.org/10.1162/isal_a_00318","title":"Safe Reinforcement Learning through Meta-learned Instincts","display_name":"Safe Reinforcement Learning through Meta-learned Instincts","publication_year":2020,"publication_date":"2020-01-01","ids":{"openalex":"https://openalex.org/W3023426586","doi":"https://doi.org/10.1162/isal_a_00318","mag":"3023426586"},"language":"en","primary_location":{"id":"doi:10.1162/isal_a_00318","is_oa":true,"landing_page_url":"https://doi.org/10.1162/isal_a_00318","pdf_url":"https://www.mitpressjournals.org/doi/pdf/10.1162/isal_a_00318","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"The 2020 Conference on Artificial Life","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://www.mitpressjournals.org/doi/pdf/10.1162/isal_a_00318","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5089907783","display_name":"Djordje Grbic","orcid":"https://orcid.org/0000-0002-4388-2539"},"institutions":[{"id":"https://openalex.org/I83467386","display_name":"IT University of Copenhagen","ror":"https://ror.org/02309jg23","country_code":"DK","type":"education","lineage":["https://openalex.org/I83467386"]}],"countries":["DK"],"is_corresponding":false,"raw_author_name":"Djordje Grbic","raw_affiliation_strings":["IT-University of Copenhagen, Denmark"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"IT-University of Copenhagen, Denmark","institution_ids":["https://openalex.org/I83467386"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5020511097","display_name":"Sebastian Risi","orcid":"https://orcid.org/0000-0003-3607-8400"},"institutions":[{"id":"https://openalex.org/I83467386","display_name":"IT University of Copenhagen","ror":"https://ror.org/02309jg23","country_code":"DK","type":"education","lineage":["https://openalex.org/I83467386"]}],"countries":["DK"],"is_corresponding":false,"raw_author_name":"Sebastian Risi","raw_affiliation_strings":["IT-UNIVERSITY OF COPENHAGEN"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"IT-UNIVERSITY OF COPENHAGEN","institution_ids":["https://openalex.org/I83467386"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I83467386"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":3,"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":0.9993000030517578,"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":0.9993000030517578,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.9977999925613403,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.9966999888420105,"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/reinforcement-learning","display_name":"Reinforcement learning","score":0.8572505712509155},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7098343968391418},{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.6255089044570923},{"id":"https://openalex.org/keywords/variety","display_name":"Variety (cybernetics)","score":0.5906703472137451},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.5471959114074707},{"id":"https://openalex.org/keywords/instinct","display_name":"Instinct","score":0.5390318036079407},{"id":"https://openalex.org/keywords/software-deployment","display_name":"Software deployment","score":0.5249668955802917},{"id":"https://openalex.org/keywords/action","display_name":"Action (physics)","score":0.49886083602905273},{"id":"https://openalex.org/keywords/meta-learning","display_name":"Meta learning (computer science)","score":0.49458953738212585},{"id":"https://openalex.org/keywords/human\u2013computer-interaction","display_name":"Human\u2013computer interaction","score":0.37060171365737915},{"id":"https://openalex.org/keywords/risk-analysis","display_name":"Risk analysis (engineering)","score":0.3338819146156311},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.3247554898262024},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.1275472640991211}],"concepts":[{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.8572505712509155},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7098343968391418},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.6255089044570923},{"id":"https://openalex.org/C136197465","wikidata":"https://www.wikidata.org/wiki/Q1729295","display_name":"Variety (cybernetics)","level":2,"score":0.5906703472137451},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5471959114074707},{"id":"https://openalex.org/C3232509","wikidata":"https://www.wikidata.org/wiki/Q18237","display_name":"Instinct","level":2,"score":0.5390318036079407},{"id":"https://openalex.org/C105339364","wikidata":"https://www.wikidata.org/wiki/Q2297740","display_name":"Software deployment","level":2,"score":0.5249668955802917},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.49886083602905273},{"id":"https://openalex.org/C2781002164","wikidata":"https://www.wikidata.org/wiki/Q6822311","display_name":"Meta learning (computer science)","level":3,"score":0.49458953738212585},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.37060171365737915},{"id":"https://openalex.org/C112930515","wikidata":"https://www.wikidata.org/wiki/Q4389547","display_name":"Risk analysis (engineering)","level":1,"score":0.3338819146156311},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3247554898262024},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.1275472640991211},{"id":"https://openalex.org/C71924100","wikidata":"https://www.wikidata.org/wiki/Q11190","display_name":"Medicine","level":0,"score":0.0},{"id":"https://openalex.org/C201995342","wikidata":"https://www.wikidata.org/wiki/Q682496","display_name":"Systems engineering","level":1,"score":0.0},{"id":"https://openalex.org/C78458016","wikidata":"https://www.wikidata.org/wiki/Q840400","display_name":"Evolutionary biology","level":1,"score":0.0},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.0},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.0},{"id":"https://openalex.org/C62520636","wikidata":"https://www.wikidata.org/wiki/Q944","display_name":"Quantum mechanics","level":1,"score":0.0},{"id":"https://openalex.org/C86803240","wikidata":"https://www.wikidata.org/wiki/Q420","display_name":"Biology","level":0,"score":0.0}],"mesh":[],"locations_count":5,"locations":[{"id":"doi:10.1162/isal_a_00318","is_oa":true,"landing_page_url":"https://doi.org/10.1162/isal_a_00318","pdf_url":"https://www.mitpressjournals.org/doi/pdf/10.1162/isal_a_00318","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"The 2020 Conference on Artificial Life","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2005.03233","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2005.03233","pdf_url":"https://arxiv.org/pdf/2005.03233","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"text"},{"id":"mag:3023426586","is_oa":true,"landing_page_url":"https://arxiv.org/pdf/2005.03233.pdf","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"arXiv (Cornell University)","raw_type":null},{"id":"pmh:oai:pure.atira.dk:publications/fefbb96e-cd02-4b75-a166-381a3295b91c","is_oa":false,"landing_page_url":"https://pure.itu.dk/portal/da/publications/fefbb96e-cd02-4b75-a166-381a3295b91c","pdf_url":null,"source":{"id":"https://openalex.org/S4377196680","display_name":"IT University Of Copenhagen (IT University of Copenhagen)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I83467386","host_organization_name":"IT University of Copenhagen","host_organization_lineage":["https://openalex.org/I83467386"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"Grbic, D & Risi, S 2020, Safe Reinforcement Learning through Meta-learned Instincts. in ALIFE 2020 : The 2020 Conference on Artificial Life. 32 edn, MIT Press, Artificial Life Conference Proceedings , pp. 183-291. https://doi.org/10.1162/isal_a_00318","raw_type":"info:eu-repo/semantics/publishedVersion"},{"id":"doi:10.48550/arxiv.2005.03233","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2005.03233","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":"doi:10.1162/isal_a_00318","is_oa":true,"landing_page_url":"https://doi.org/10.1162/isal_a_00318","pdf_url":"https://www.mitpressjournals.org/doi/pdf/10.1162/isal_a_00318","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"The 2020 Conference on Artificial Life","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":30,"referenced_works":["https://openalex.org/W1518931405","https://openalex.org/W1522301498","https://openalex.org/W1677182931","https://openalex.org/W1845972764","https://openalex.org/W2089217417","https://openalex.org/W2096952131","https://openalex.org/W2119717200","https://openalex.org/W2122053572","https://openalex.org/W2288565641","https://openalex.org/W2555811267","https://openalex.org/W2580175322","https://openalex.org/W2604763608","https://openalex.org/W2605353576","https://openalex.org/W2736601468","https://openalex.org/W2749928749","https://openalex.org/W2760552731","https://openalex.org/W2778749116","https://openalex.org/W2784465508","https://openalex.org/W2808809512","https://openalex.org/W2889696653","https://openalex.org/W2922874720","https://openalex.org/W2949608212","https://openalex.org/W2954989419","https://openalex.org/W2955766841","https://openalex.org/W2962258836","https://openalex.org/W2962803570","https://openalex.org/W2962991582","https://openalex.org/W2963293747","https://openalex.org/W2963575966","https://openalex.org/W2994180726"],"related_works":["https://openalex.org/W3046241732","https://openalex.org/W2588415587","https://openalex.org/W2953473809","https://openalex.org/W161119602","https://openalex.org/W2765222421","https://openalex.org/W2647735349","https://openalex.org/W1830953589","https://openalex.org/W2899629811","https://openalex.org/W2808037395","https://openalex.org/W3037804676","https://openalex.org/W2462355218","https://openalex.org/W3005850366","https://openalex.org/W2532299421","https://openalex.org/W3169675513","https://openalex.org/W2783268317","https://openalex.org/W928452086","https://openalex.org/W3201551014","https://openalex.org/W2980389916","https://openalex.org/W3104013016","https://openalex.org/W2059841094"],"abstract_inverted_index":{"An":[0],"important":[1],"goal":[2],"in":[3,45,56,63,84,162,220],"reinforcement":[4,47,124],"learning":[5,125],"is":[6,33,50,118,208],"to":[7,14,24,40,81,101,167,169,183,185,187],"create":[8],"agents":[9,31,100,182],"that":[10,20,72,98,201],"can":[11,140],"quickly":[12],"adapt":[13],"new":[15,42,171,188,211],"goals":[16],"while":[17,107],"avoiding":[18,108],"situations":[19],"might":[21],"cause":[22],"damage":[23],"themselves":[25],"or":[26],"their":[27,105],"environments.":[28],"One":[29],"way":[30],"learn":[32,103,168,184],"through":[34,123],"exploration":[35,49],"mechanisms,":[36],"which":[37,131,163,216],"are":[38],"needed":[39],"discover":[41],"policies.":[43],"However,":[44],"deep":[46],"learning,":[48],"normally":[51],"done":[52],"by":[53,77],"injecting":[54],"noise":[55],"the":[57,69,73,78,85,113,116,136,142,146,164,195],"action":[58],"space.":[59],"While":[60],"performing":[61],"well":[62],"many":[64],"domains,":[65],"this":[66,221],"setup":[67],"has":[68,166],"inherent":[70],"risk":[71],"noisy":[74,143],"actions":[75],"performed":[76],"agent":[79,165],"lead":[80],"unsafe":[82],"states":[83],"environment.":[86],"Here":[87],"we":[88],"introduce":[89],"a":[90,119,154,209,224],"novel":[91],"approach":[92,117,170,212],"called":[93],"Meta-Learned":[94],"Instinctual":[95],"Networks":[96],"(MLIN)":[97],"allows":[99,181],"safely":[102],"during":[104,135,173],"lifetime":[106,138],"potentially":[109],"hazardous":[110],"states.":[111],"At":[112],"core":[114],"of":[115,145,194,226],"plastic":[120,147],"network":[121,207],"trained":[122],"and":[126,180],"an":[127,205],"evolved":[128],"\"instinctual\"":[129],"network,":[130],"does":[132],"not":[133],"change":[134],"agent's":[137],"but":[139],"modulate":[141],"output":[144],"network.":[148],"We":[149],"test":[150],"our":[151],"idea":[152],"on":[153,223],"simple":[155],"2D":[156],"navigation":[157],"task":[158],"with":[159,192,204],"no-go":[160,196],"zones,":[161],"targets":[172,189],"deployment.":[174],"MLIN":[175],"outperforms":[176],"standard":[177],"meta-trained":[178],"networks":[179],"navigate":[186],"without":[190],"colliding":[191],"any":[193],"zones.":[197],"These":[198],"results":[199],"suggest":[200],"meta-learning":[202],"augmented":[203],"instinctual":[206],"promising":[210],"for":[213],"safe":[214],"AI,":[215],"may":[217],"enable":[218],"progress":[219],"area":[222],"variety":[225],"different":[227],"domains.":[228]},"counts_by_year":[{"year":2023,"cited_by_count":1},{"year":2021,"cited_by_count":1},{"year":2020,"cited_by_count":1}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}