{"id":"https://openalex.org/W3196816907","doi":"https://doi.org/10.48550/arxiv.2108.12001","title":"Understanding the Logit Distributions of Adversarially-Trained Deep Neural Networks","display_name":"Understanding the Logit Distributions of Adversarially-Trained Deep Neural Networks","publication_year":2021,"publication_date":"2021-08-26","ids":{"openalex":"https://openalex.org/W3196816907","doi":"https://doi.org/10.48550/arxiv.2108.12001","mag":"3196816907"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2108.12001","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2108.12001","pdf_url":"https://arxiv.org/pdf/2108.12001","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/2108.12001","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5073877456","display_name":"Landan Seguin","orcid":"https://orcid.org/0000-0002-8020-5887","observed_orcids":["https://orcid.org/0000-0002-8020-5887"]},"institutions":[{"id":"https://openalex.org/I78577930","display_name":"Columbia University","ror":"https://ror.org/00hj8s172","country_code":"US","type":"education","lineage":["https://openalex.org/I78577930"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Seguin, Landan","raw_affiliation_strings":["Columbia University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Columbia University","institution_ids":["https://openalex.org/I78577930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5024372333","display_name":"Anthony Ndirango","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I1343180700","display_name":"Intel (United States)","ror":"https://ror.org/01ek73717","country_code":"US","type":"company","lineage":["https://openalex.org/I1343180700"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Ndirango, Anthony","raw_affiliation_strings":["Intel Labs","Intel,,,,,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Intel Labs","institution_ids":["https://openalex.org/I1343180700"]},{"raw_affiliation_string":"Intel,,,,,","institution_ids":["https://openalex.org/I1343180700"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5039722376","display_name":"Neeli Mishra","orcid":"https://orcid.org/0000-0002-2946-4445","observed_orcids":["https://orcid.org/0000-0002-2946-4445"]},"institutions":[{"id":"https://openalex.org/I78577930","display_name":"Columbia University","ror":"https://ror.org/00hj8s172","country_code":"US","type":"education","lineage":["https://openalex.org/I78577930"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Mishra, Neeli","raw_affiliation_strings":["Columbia University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Columbia University","institution_ids":["https://openalex.org/I78577930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5016533438","display_name":"SueYeon Chung","orcid":"https://orcid.org/0000-0001-5726-4865","observed_orcids":["https://orcid.org/0000-0001-5726-4865"]},"institutions":[{"id":"https://openalex.org/I78577930","display_name":"Columbia University","ror":"https://ror.org/00hj8s172","country_code":"US","type":"education","lineage":["https://openalex.org/I78577930"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Chung, SueYeon","raw_affiliation_strings":["Columbia University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Columbia University","institution_ids":["https://openalex.org/I78577930"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5088664821","display_name":"Tyler Lee","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I1343180700","display_name":"Intel (United States)","ror":"https://ror.org/01ek73717","country_code":"US","type":"company","lineage":["https://openalex.org/I1343180700"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Lee, Tyler","raw_affiliation_strings":["Intel Labs","Intel,,,,,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Intel Labs","institution_ids":["https://openalex.org/I1343180700"]},{"raw_affiliation_string":"Intel,,,,,","institution_ids":["https://openalex.org/I1343180700"]}]}],"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":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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":1.0,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":1.0,"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/T11424","display_name":"Security and Verification in Computing","score":0.0,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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/T12535","display_name":"Machine Learning and Data Classification","score":0.0,"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/adversarial-training","display_name":"adversarial training","score":0.9950000047683716},{"id":"https://openalex.org/keywords/adversarial-robustness","display_name":"adversarial robustness","score":0.9850000143051147},{"id":"https://openalex.org/keywords/adversarial-defense","display_name":"adversarial defense","score":0.671999990940094},{"id":"https://openalex.org/keywords/knowledge-distillation","display_name":"knowledge distillation","score":0.9419999718666077},{"id":"https://openalex.org/keywords/confidence-distributions","display_name":"confidence distributions","score":0.44200000166893005},{"id":"https://openalex.org/keywords/deep-neural-network","display_name":"deep neural network","score":0.6290000081062317}],"concepts":[{"id":"https://openalex.org/C140331021","wikidata":"https://www.wikidata.org/wiki/Q1868104","display_name":"Logit","level":2,"score":0.744280219078064},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.6517289876937866},{"id":"https://openalex.org/C149782125","wikidata":"https://www.wikidata.org/wiki/Q160039","display_name":"Econometrics","level":1,"score":0.5039026141166687},{"id":"https://openalex.org/C151956035","wikidata":"https://www.wikidata.org/wiki/Q1132755","display_name":"Logistic regression","level":2,"score":0.5017566680908203},{"id":"https://openalex.org/C2984842247","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep neural networks","level":3,"score":0.47803840041160583},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4681195318698883},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.4106411039829254},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.3655429482460022},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.32908108830451965},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.28632456064224243}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2108.12001","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2108.12001","pdf_url":"https://arxiv.org/pdf/2108.12001","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.2108.12001","doi":"https://doi.org/10.48550/arxiv.2108.12001","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2108.12001","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":"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"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2108.12001","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2108.12001","pdf_url":"https://arxiv.org/pdf/2108.12001","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/W3196816907.pdf","grobid_xml":"https://content.openalex.org/works/W3196816907.grobid-xml"},"referenced_works_count":41,"referenced_works":["https://openalex.org/W1821462560","https://openalex.org/W2021581601","https://openalex.org/W2174868984","https://openalex.org/W2183341477","https://openalex.org/W2194775991","https://openalex.org/W2243397390","https://openalex.org/W2516574342","https://openalex.org/W2552767274","https://openalex.org/W2765233338","https://openalex.org/W2766572821","https://openalex.org/W2787708942","https://openalex.org/W2791953061","https://openalex.org/W2803546054","https://openalex.org/W2895097814","https://openalex.org/W2898152545","https://openalex.org/W2908392948","https://openalex.org/W2913266441","https://openalex.org/W2945289329","https://openalex.org/W2948447326","https://openalex.org/W2952610664","https://openalex.org/W2963207607","https://openalex.org/W2964153729","https://openalex.org/W2964222566","https://openalex.org/W2964253222","https://openalex.org/W2970421120","https://openalex.org/W2970971581","https://openalex.org/W2996344901","https://openalex.org/W2997560666","https://openalex.org/W3004639598","https://openalex.org/W3004809259","https://openalex.org/W3006834354","https://openalex.org/W3007305010","https://openalex.org/W3008105014","https://openalex.org/W3009542902","https://openalex.org/W3016970897","https://openalex.org/W3034368386","https://openalex.org/W3085383590","https://openalex.org/W3093019879","https://openalex.org/W3103984627","https://openalex.org/W3116515605","https://openalex.org/W3118608800"],"related_works":["https://openalex.org/W4280506080","https://openalex.org/W2948447326","https://openalex.org/W2884821828","https://openalex.org/W4386302519","https://openalex.org/W3006335454","https://openalex.org/W7133265066","https://openalex.org/W3027735248","https://openalex.org/W2970682262","https://openalex.org/W2767075075","https://openalex.org/W4321019893"],"abstract_inverted_index":{"Adversarial":[0],"defenses":[1],"train":[2],"deep":[3],"neural":[4],"networks":[5],"to":[6,9,95,182,185,218],"be":[7,183],"invariant":[8],"the":[10,42,82,106,116,119,124,129,170,190,198,201],"input":[11,65,213],"perturbations":[12,66,214],"from":[13],"adversarial":[14,24,31,34,40,78,97,109,210],"attacks.":[15],"Almost":[16],"all":[17],"defense":[18],"strategies":[19],"achieve":[20],"this":[21],"invariance":[22],"through":[23],"training":[25,27,35,79,110],"i.e.":[26],"on":[28,61,137,153,200],"inputs":[29],"with":[30,169],"perturbations.":[32],"Although":[33],"is":[36,75],"successful":[37],"at":[38],"mitigating":[39],"attacks,":[41],"behavioral":[43],"differences":[44,165],"between":[45,128],"adversarially-trained":[46],"(AT)":[47],"models":[48,51,150],"and":[49,123,132,148,196,223],"standard":[50,149],"are":[52,136,156],"still":[53],"poorly":[54],"understood.":[55],"Motivated":[56],"by":[57,67,80,166,188],"a":[58,102,216],"recent":[59],"study":[60],"learning":[62,96,177,186,208],"robustness":[63,187,211],"without":[64,212],"distilling":[68],"an":[69],"AT":[70,87,141,147],"model,":[71],"we":[72,100,144,175],"explore":[73],"what":[74],"learned":[76],"during":[77,194],"analyzing":[81],"distribution":[83],"of":[84,115],"logits":[85],"in":[86],"models.":[88,142],"We":[89],"identify":[90],"three":[91],"logit":[92,117,121,130,192],"characteristics":[93,114],"essential":[94,184],"robustness.":[98,203],"First,":[99],"provide":[101],"theoretical":[103],"justification":[104],"for":[105,140],"finding":[107],"that":[108,146,207,227],"shrinks":[111],"two":[112],"important":[113],"distribution:":[118],"max":[120,131],"values":[122],"\"logit":[125],"gaps\"":[126],"(difference":[127],"next":[133],"largest":[134,171],"values)":[135],"average":[138],"lower":[139],"Second,":[143],"show":[145],"differ":[151],"significantly":[152],"which":[154],"samples":[155,168],"high":[157],"or":[158],"low":[159],"confidence,":[160],"then":[161],"illustrate":[162],"clear":[163],"qualitative":[164],"visualizing":[167],"confidence":[172],"difference.":[173],"Finally,":[174],"find":[176],"information":[178,193],"about":[179],"incorrect":[180,224],"classes":[181],"manipulating":[189],"non-max":[191],"distillation":[195],"measuring":[197],"impact":[199],"student's":[202],"Our":[204],"results":[205],"indicate":[206],"some":[209],"requires":[215],"model":[217],"learn":[219],"specific":[220],"sample-wise":[221],"confidences":[222],"class":[225],"orderings":[226],"follow":[228],"complex":[229],"distributions.":[230]},"counts_by_year":[{"year":2025,"cited_by_count":2},{"year":2024,"cited_by_count":1}],"updated_date":"2026-11-05T12:25:35.769452","created_date":"2021-09-13T00:00:00"}