{"id":"https://openalex.org/W4416932001","doi":"https://doi.org/10.48550/arxiv.2509.25686","title":"EXP-CAM: Explanation Generation and Circuit Discovery Using Classifier Activation Matching","display_name":"EXP-CAM: Explanation Generation and Circuit Discovery Using Classifier Activation Matching","publication_year":2025,"publication_date":"2025-09-30","ids":{"openalex":"https://openalex.org/W4416932001","doi":"https://doi.org/10.48550/arxiv.2509.25686"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2509.25686","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2509.25686","pdf_url":"https://arxiv.org/pdf/2509.25686","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","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2509.25686","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5093967906","display_name":"Pirzada Suhail","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I162827531","display_name":"Indian Institute of Technology Bombay","ror":"https://ror.org/02qyf5152","country_code":"IN","type":"education","lineage":["https://openalex.org/I162827531"]}],"countries":["IN"],"is_corresponding":false,"raw_author_name":"Suhail, Pirzada","raw_affiliation_strings":["Department of Electrical Engineering IIT Bombay"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Electrical Engineering IIT Bombay","institution_ids":["https://openalex.org/I162827531"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101735364","display_name":"Aditya Anand","orcid":"https://orcid.org/0000-0001-6615-7463","observed_orcids":["https://orcid.org/0000-0001-6615-7463"]},"institutions":[{"id":"https://openalex.org/I162827531","display_name":"Indian Institute of Technology Bombay","ror":"https://ror.org/02qyf5152","country_code":"IN","type":"education","lineage":["https://openalex.org/I162827531"]}],"countries":["IN"],"is_corresponding":false,"raw_author_name":"Anand, Aditya","raw_affiliation_strings":["Department of Electrical Engineering IIT Bombay"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Electrical Engineering IIT Bombay","institution_ids":["https://openalex.org/I162827531"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5087629613","display_name":"Amit Sethi","orcid":"https://orcid.org/0000-0002-8634-1804","observed_orcids":["https://orcid.org/0000-0002-8634-1804"]},"institutions":[{"id":"https://openalex.org/I162827531","display_name":"Indian Institute of Technology Bombay","ror":"https://ror.org/02qyf5152","country_code":"IN","type":"education","lineage":["https://openalex.org/I162827531"]}],"countries":["IN"],"is_corresponding":false,"raw_author_name":"Sethi, Amit","raw_affiliation_strings":["Department of Electrical Engineering IIT Bombay"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Electrical Engineering IIT Bombay","institution_ids":["https://openalex.org/I162827531"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I162827531"],"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/T12026","display_name":"Explainable Artificial Intelligence (XAI)","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/T12026","display_name":"Explainable Artificial Intelligence (XAI)","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/T10036","display_name":"Advanced Neural Network Applications","score":0.0,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/T10743","display_name":"Software Testing and Debugging Techniques","score":0.0,"subfield":{"id":"https://openalex.org/subfields/1712","display_name":"Software"},"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/explanation-generation","display_name":"explanation generation","score":0.7020000219345093},{"id":"https://openalex.org/keywords/circuit-discovery","display_name":"circuit discovery","score":0.7710000276565552},{"id":"https://openalex.org/keywords/minimal-explanations","display_name":"minimal explanations","score":0.8659999966621399},{"id":"https://openalex.org/keywords/image-classifiers","display_name":"image classifiers","score":0.4659999907016754},{"id":"https://openalex.org/keywords/autoencoder","display_name":"autoencoder","score":0.7379999756813049},{"id":"https://openalex.org/keywords/binary-masks","display_name":"binary masks","score":0.5600000023841858},{"id":"https://openalex.org/keywords/explanation-faithfulness","display_name":"explanation fidelity","score":0.746999979019165},{"id":"https://openalex.org/keywords/mechanistic-interpretability","display_name":"mechanistic interpretability","score":0.7400000095367432},{"id":"https://openalex.org/keywords/gradient-based-explanation","display_name":"gradient-based explanation","score":0.30300000309944153}],"concepts":[{"id":"https://openalex.org/C95623464","wikidata":"https://www.wikidata.org/wiki/Q1096149","display_name":"Classifier (UML)","level":2,"score":0.741100013256073},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7006999850273132},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6261000037193298},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.5461000204086304},{"id":"https://openalex.org/C3309909","wikidata":"https://www.wikidata.org/wiki/Q864155","display_name":"Binary decision diagram","level":2,"score":0.5310999751091003},{"id":"https://openalex.org/C66905080","wikidata":"https://www.wikidata.org/wiki/Q17005494","display_name":"Binary classification","level":3,"score":0.47690001130104065},{"id":"https://openalex.org/C177769412","wikidata":"https://www.wikidata.org/wiki/Q278090","display_name":"Prior probability","level":3,"score":0.44670000672340393},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.4307999908924103},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.42149999737739563},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.40459999442100525},{"id":"https://openalex.org/C48372109","wikidata":"https://www.wikidata.org/wiki/Q3913","display_name":"Binary number","level":2,"score":0.3885999917984009},{"id":"https://openalex.org/C84839998","wikidata":"https://www.wikidata.org/wiki/Q5249245","display_name":"Decision rule","level":2,"score":0.3160000145435333},{"id":"https://openalex.org/C42023084","wikidata":"https://www.wikidata.org/wiki/Q5249231","display_name":"Decision boundary","level":3,"score":0.30630001425743103},{"id":"https://openalex.org/C28901747","wikidata":"https://www.wikidata.org/wiki/Q177571","display_name":"Decision theory","level":2,"score":0.3012999892234802},{"id":"https://openalex.org/C2781067378","wikidata":"https://www.wikidata.org/wiki/Q17027399","display_name":"Interpretability","level":2,"score":0.30090001225471497},{"id":"https://openalex.org/C46686674","wikidata":"https://www.wikidata.org/wiki/Q466303","display_name":"Boosting (machine learning)","level":2,"score":0.2913999855518341},{"id":"https://openalex.org/C52622490","wikidata":"https://www.wikidata.org/wiki/Q1026626","display_name":"Feature extraction","level":2,"score":0.28450000286102295},{"id":"https://openalex.org/C75294576","wikidata":"https://www.wikidata.org/wiki/Q5165192","display_name":"Contextual image classification","level":3,"score":0.28299999237060547},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.2777999937534332},{"id":"https://openalex.org/C2983787585","wikidata":"https://www.wikidata.org/wiki/Q93586","display_name":"Feature matching","level":3,"score":0.2752000093460083},{"id":"https://openalex.org/C75553542","wikidata":"https://www.wikidata.org/wiki/Q178161","display_name":"A priori and a posteriori","level":2,"score":0.26829999685287476},{"id":"https://openalex.org/C83665646","wikidata":"https://www.wikidata.org/wiki/Q42139305","display_name":"Feature vector","level":2,"score":0.2597000002861023},{"id":"https://openalex.org/C2776036281","wikidata":"https://www.wikidata.org/wiki/Q48769818","display_name":"Constraint (computer-aided design)","level":2,"score":0.25099998712539673},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.25040000677108765}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2509.25686","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2509.25686","pdf_url":"https://arxiv.org/pdf/2509.25686","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":"doi:10.48550/arxiv.2509.25686","doi":"https://doi.org/10.48550/arxiv.2509.25686","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2509.25686","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:2509.25686","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2509.25686","pdf_url":"https://arxiv.org/pdf/2509.25686","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":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4416932001.pdf","grobid_xml":"https://content.openalex.org/works/W4416932001.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7083834007","https://openalex.org/W3037211332","https://openalex.org/W7215040426","https://openalex.org/W4396759605","https://openalex.org/W3015764181","https://openalex.org/W7131413355","https://openalex.org/W7154571049","https://openalex.org/W3166484217","https://openalex.org/W4320036915","https://openalex.org/W4392209659"],"abstract_inverted_index":{"Machine":[0],"learning":[1],"models,":[2],"by":[3,204,214],"virtue":[4],"of":[5,11,21,69,125,236],"training,":[6],"learn":[7],"a":[8,27,110,154,180,198,225,233],"large":[9],"repertoire":[10],"decision":[12,79,98],"rules":[13,35],"for":[14,66],"any":[15,19],"given":[16],"input,":[17],"and":[18,40,51,63,84,103,150,190,196,211,232],"one":[20],"these":[22,222],"may":[23],"suffice":[24],"to":[25,38,88,113,119,168],"justify":[26],"prediction.":[28],"However,":[29],"in":[30],"high-dimensional":[31],"input":[32],"spaces,":[33],"such":[34],"are":[36,81,100],"difficult":[37],"identify":[39,89,193],"interpret.":[41],"In":[42,174],"this":[43,175],"paper,":[44],"we":[45,108,177,192],"introduce":[46,179],"EXP-CAM:":[47],"an":[48,126],"explanation":[49],"generation":[50],"circuit":[52,181],"discovery":[53],"approach":[54],"using":[55,185],"Classifier":[56],"Activation":[57],"Matching.":[58],"EXP-CAM":[59],"can":[60],"generate":[61],"minimal":[62,90,161,229],"faithful":[64],"explanations":[65,91,162,231],"the":[67,77,96,121,143,171,186,237],"decisions":[68],"pre-trained":[70],"image":[71,127],"classifiers":[72],"that":[73,92,117,147,157],"not":[74,93],"only":[75,94],"preserve":[76,95],"model's":[78,97],"but":[80,99],"also":[82,101,165,178],"concise":[83,102],"human-readable.":[85,104],"We":[86],"aim":[87],"To":[105],"achieve":[106],"this,":[107],"train":[109],"lightweight":[111],"auto-encoder":[112],"produce":[114],"binary":[115],"masks":[116],"learns":[118],"highlight":[120],"decision-wise":[122],"critical":[123],"regions":[124],"while":[128],"discarding":[129],"irrelevant":[130],"background.":[131],"The":[132,160],"training":[133],"objective":[134],"integrates":[135],"activation":[136,210],"alignment":[137],"across":[138],"multiple":[139],"layers,":[140],"consistency":[141],"at":[142],"output":[144],"label,":[145],"priors":[146],"encourage":[148],"sparsity,":[149],"compactness,":[151],"along":[152],"with":[153],"robustness":[155],"constraint":[156],"enforces":[158],"faithfulness.":[159],"so":[163],"generated":[164],"lead":[166],"us":[167],"mechanistically":[169],"interpreting":[170],"model":[172,241],"internals.":[173],"regard":[176],"readout":[182],"procedure":[183],"wherein":[184],"explanation's":[187],"forward":[188],"pass":[189],"gradients,":[191],"active":[194],"channels":[195],"construct":[197],"channel-level":[199],"graph,":[200],"scoring":[201],"inter-layer":[202],"edges":[203],"ingress":[205],"weight":[206,216],"magnitude":[207,217],"times":[208,218],"source":[209],"feature-to-class":[212],"links":[213],"classifier":[215],"feature":[219],"activation.":[220],"Together,":[221],"contributions":[223],"provide":[224],"practical":[226],"bridge":[227],"between":[228],"input-level":[230],"mechanistic":[234],"understanding":[235],"internal":[238],"computations":[239],"driving":[240],"decisions.":[242]},"counts_by_year":[],"updated_date":"2026-09-30T09:27:27.532063","created_date":"2025-10-10T00:00:00"}