{"id":"https://openalex.org/W4282813352","doi":"https://doi.org/10.48550/arxiv.2206.04920","title":"Fisher SAM: Information Geometry and Sharpness Aware Minimisation","display_name":"Fisher SAM: Information Geometry and Sharpness Aware Minimisation","publication_year":2022,"publication_date":"2022-06-10","ids":{"openalex":"https://openalex.org/W4282813352","doi":"https://doi.org/10.48550/arxiv.2206.04920"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2206.04920","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2206.04920","pdf_url":"https://arxiv.org/pdf/2206.04920","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2206.04920","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5100670947","display_name":"Minyoung Kim","orcid":"https://orcid.org/0000-0003-4273-7118","observed_orcids":["https://orcid.org/0000-0003-4273-7118","https://orcid.org/0009-0008-5288-5337"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kim, Minyoung","raw_affiliation_strings":["Samsung AI Center , Cambridge , UK"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Samsung AI Center , Cambridge , UK","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100384557","display_name":"Da Li","orcid":"https://orcid.org/0000-0003-4188-5397","observed_orcids":["https://orcid.org/0000-0003-4188-5397"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Da","raw_affiliation_strings":["Samsung AI Center , Cambridge , UK"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Samsung AI Center , Cambridge , UK","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5032445207","display_name":"Shell Xu Hu","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hu, Shell Xu","raw_affiliation_strings":["Samsung AI Center , Cambridge , UK"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Samsung AI Center , Cambridge , UK","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5087823932","display_name":"Timothy M. Hospedales","orcid":"https://orcid.org/0000-0003-4867-7486","observed_orcids":["https://orcid.org/0000-0003-4867-7486"]},"institutions":[{"id":"https://openalex.org/I98677209","display_name":"University of Edinburgh","ror":"https://ror.org/01nrxwf90","country_code":"GB","type":"education","lineage":["https://openalex.org/I98677209"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Hospedales, Timothy M.","raw_affiliation_strings":["Samsung AI Center , Cambridge , UK","University of Edinburgh "],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Samsung AI Center , Cambridge , UK","institution_ids":[]},{"raw_affiliation_string":"University of Edinburgh ","institution_ids":["https://openalex.org/I98677209"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":16,"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.9765999913215637,"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.9765999913215637,"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/T12535","display_name":"Machine Learning and Data Classification","score":0.01119999960064888,"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":0.0032999999821186066,"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/fisher-information","display_name":"Fisher information","score":0.8930000066757202},{"id":"https://openalex.org/keywords/information-geometry","display_name":"information geometry","score":0.9559999704360962},{"id":"https://openalex.org/keywords/statistical-manifolds","display_name":"statistical manifolds","score":0.9490000009536743},{"id":"https://openalex.org/keywords/flat-minima","display_name":"flat minima","score":0.9240000247955322},{"id":"https://openalex.org/keywords/generalization","display_name":"generalization","score":0.7760000228881836},{"id":"https://openalex.org/keywords/robustness","display_name":"robustness","score":0.9279999732971191},{"id":"https://openalex.org/keywords/neural-network-training","display_name":"neural network training","score":0.5799999833106995},{"id":"https://openalex.org/keywords/neighborhood-structure","display_name":"neighborhood structure","score":0.37299999594688416}],"concepts":[{"id":"https://openalex.org/C29406490","wikidata":"https://www.wikidata.org/wiki/Q1420659","display_name":"Fisher information","level":2,"score":0.6436810493469238},{"id":"https://openalex.org/C109546454","wikidata":"https://www.wikidata.org/wiki/Q3798604","display_name":"Information geometry","level":4,"score":0.6352478265762329},{"id":"https://openalex.org/C129782007","wikidata":"https://www.wikidata.org/wiki/Q162886","display_name":"Euclidean geometry","level":2,"score":0.6348323225975037},{"id":"https://openalex.org/C186633575","wikidata":"https://www.wikidata.org/wiki/Q845060","display_name":"Maxima and minima","level":2,"score":0.5752378106117249},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.547416090965271},{"id":"https://openalex.org/C122041747","wikidata":"https://www.wikidata.org/wiki/Q838611","display_name":"Ball (mathematics)","level":2,"score":0.5223482847213745},{"id":"https://openalex.org/C186450821","wikidata":"https://www.wikidata.org/wiki/Q17295","display_name":"Euclidean space","level":2,"score":0.4902415871620178},{"id":"https://openalex.org/C73586568","wikidata":"https://www.wikidata.org/wiki/Q2600211","display_name":"Parameter space","level":2,"score":0.47834035754203796},{"id":"https://openalex.org/C18294631","wikidata":"https://www.wikidata.org/wiki/Q2621185","display_name":"Location parameter","level":3,"score":0.4659707546234131},{"id":"https://openalex.org/C205711294","wikidata":"https://www.wikidata.org/wiki/Q176953","display_name":"Rendering (computer graphics)","level":2,"score":0.43202900886535645},{"id":"https://openalex.org/C86941820","wikidata":"https://www.wikidata.org/wiki/Q6865391","display_name":"Minimisation (clinical trials)","level":2,"score":0.42543739080429077},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.4174405336380005},{"id":"https://openalex.org/C207798031","wikidata":"https://www.wikidata.org/wiki/Q8563425","display_name":"Fisher kernel","level":5,"score":0.41618895530700684},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.4119942784309387},{"id":"https://openalex.org/C120174047","wikidata":"https://www.wikidata.org/wiki/Q847073","display_name":"Euclidean distance","level":2,"score":0.3814294934272766},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.371018648147583},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.35994553565979004},{"id":"https://openalex.org/C2524010","wikidata":"https://www.wikidata.org/wiki/Q8087","display_name":"Geometry","level":1,"score":0.3515653908252716},{"id":"https://openalex.org/C149441793","wikidata":"https://www.wikidata.org/wiki/Q200726","display_name":"Probability distribution","level":2,"score":0.3012867569923401},{"id":"https://openalex.org/C202444582","wikidata":"https://www.wikidata.org/wiki/Q837863","display_name":"Pure mathematics","level":1,"score":0.18159928917884827},{"id":"https://openalex.org/C134306372","wikidata":"https://www.wikidata.org/wiki/Q7754","display_name":"Mathematical analysis","level":1,"score":0.18132942914962769},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.18070188164710999},{"id":"https://openalex.org/C195065555","wikidata":"https://www.wikidata.org/wiki/Q214881","display_name":"Curvature","level":2,"score":0.13920271396636963},{"id":"https://openalex.org/C31510193","wikidata":"https://www.wikidata.org/wiki/Q1192553","display_name":"Facial recognition system","level":3,"score":0.0},{"id":"https://openalex.org/C181367576","wikidata":"https://www.wikidata.org/wiki/Q6394184","display_name":"Kernel Fisher discriminant analysis","level":4,"score":0.0},{"id":"https://openalex.org/C104317684","wikidata":"https://www.wikidata.org/wiki/Q7187","display_name":"Gene","level":2,"score":0.0},{"id":"https://openalex.org/C12520029","wikidata":"https://www.wikidata.org/wiki/Q1147161","display_name":"Scalar curvature","level":3,"score":0.0},{"id":"https://openalex.org/C185592680","wikidata":"https://www.wikidata.org/wiki/Q2329","display_name":"Chemistry","level":0,"score":0.0},{"id":"https://openalex.org/C55493867","wikidata":"https://www.wikidata.org/wiki/Q7094","display_name":"Biochemistry","level":1,"score":0.0}],"mesh":[],"locations_count":3,"locations":[{"id":"pmh:oai:arXiv.org:2206.04920","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2206.04920","pdf_url":"https://arxiv.org/pdf/2206.04920","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"pmh:oai:pure.ed.ac.uk:openaire/6741d131-8775-421d-a828-3b1e64bba420","doi":null,"is_oa":true,"landing_page_url":"https://www.research.ed.ac.uk/en/publications/6741d131-8775-421d-a828-3b1e64bba420","pdf_url":null,"source":{"id":"https://openalex.org/S4306400321","display_name":"Edinburgh Research Explorer (University of Edinburgh)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I98677209","host_organization_name":"University of Edinburgh","host_organization_lineage":["https://openalex.org/I98677209"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"Kim, M, Li, D, Hu, S X & Hospedales, T M 2022, Fisher SAM: Information Geometry and Sharpness Aware Minimisation. in K Chaudhuri, S Jegelka, L Song, C Szepesvari, G Niu & S Sabato (eds), Proceedings of the 39th International Conference on Machine Learning. Proceedings of Machine Learning Research, vol. 162, pp. 11148-11161, 39th International Conference on Machine Learning, Baltimore, Maryland, United States, 17/07/22. < https://proceedings.mlr.press/v162/kim22f.html >","raw_type":"info:eu-repo/semantics/conferenceObject"},{"id":"doi:10.48550/arxiv.2206.04920","doi":"https://doi.org/10.48550/arxiv.2206.04920","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2206.04920","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:2206.04920","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2206.04920","pdf_url":"https://arxiv.org/pdf/2206.04920","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":"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/W4282813352.pdf","grobid_xml":"https://content.openalex.org/works/W4282813352.grobid-xml"},"referenced_works_count":4,"referenced_works":["https://openalex.org/W2518108298","https://openalex.org/W2746314669","https://openalex.org/W2774423163","https://openalex.org/W3086499488"],"related_works":["https://openalex.org/W7212816749","https://openalex.org/W4417315113","https://openalex.org/W4417451972","https://openalex.org/W7133188468","https://openalex.org/W7165355408","https://openalex.org/W7221885657","https://openalex.org/W3204181204","https://openalex.org/W7133245195","https://openalex.org/W7160905546","https://openalex.org/W4367626110"],"abstract_inverted_index":{"Recent":[0],"sharpness-aware":[1],"minimisation":[2],"(SAM)":[3],"is":[4,11,153],"known":[5],"to":[6,45,116,144],"find":[7],"flat":[8],"minima":[9],"which":[10,49,152],"beneficial":[12],"for":[13,56],"better":[14],"generalisation":[15],"with":[16,97,170],"improved":[17,188],"robustness.":[18],"SAM":[19,127,162,194],"essentially":[20],"modifies":[21],"the":[22,27,32,36,42,47,70,80,84,90,101,117,121,130,145,148,165,171,174,183,191],"loss":[23,29,54,132],"function":[24],"by":[25,100,155],"reporting":[26],"maximum":[28],"value":[30,133],"within":[31],"small":[33],"neighborhood":[34,112,184],"around":[35],"current":[37],"iterate.":[38],"However,":[39],"it":[40],"uses":[41],"Euclidean":[43,95,166],"ball":[44,167],"define":[46],"neighborhood,":[48,91],"can":[50],"be":[51,179],"inaccurate":[52],"since":[53],"functions":[55],"neural":[57],"networks":[58],"are":[59],"typically":[60],"defined":[61],"over":[62],"probability":[63],"distributions":[64],"(e.g.,":[65],"class":[66],"predictive":[67],"probabilities),":[68],"rendering":[69],"parameter":[71,86,149,175],"space":[72,87,150],"non":[73],"Euclidean.":[74],"In":[75],"this":[76],"paper":[77],"we":[78],"consider":[79],"information":[81],"geometry":[82],"of":[83,120,147,173,190],"model":[85],"when":[88],"defining":[89],"namely":[92],"replacing":[93],"SAM's":[94],"balls":[96],"ellipsoids":[98],"induced":[99],"Fisher":[102,107,157,193],"information.":[103],"Our":[104],"approach,":[105],"dubbed":[106],"SAM,":[108],"defines":[109],"more":[110],"accurate":[111],"structures":[113],"that":[114],"conform":[115],"intrinsic":[118],"metric":[119],"underlying":[122],"statistical":[123],"manifold.":[124],"For":[125],"instance,":[126],"may":[128],"probe":[129],"worst-case":[131],"at":[134],"either":[135],"a":[136],"too":[137],"nearby":[138],"or":[139],"inappropriately":[140],"distant":[141],"point":[142],"due":[143],"ignorance":[146],"geometry,":[151],"avoided":[154],"our":[156],"SAM.":[158],"Another":[159],"recent":[160],"Adaptive":[161],"approach":[163],"stretches/shrinks":[164],"in":[168],"accordance":[169],"scale":[172],"magnitudes.":[176],"This":[177],"might":[178],"dangerous,":[180],"potentially":[181],"destroying":[182],"structure.":[185],"We":[186],"demonstrate":[187],"performance":[189],"proposed":[192],"on":[195],"several":[196],"benchmark":[197],"datasets/tasks.":[198]},"counts_by_year":[{"year":2025,"cited_by_count":3},{"year":2024,"cited_by_count":7},{"year":2023,"cited_by_count":6}],"updated_date":"2026-11-10T11:39:43.021295","created_date":"2022-06-14T00:00:00"}