{"id":"https://openalex.org/W7157604428","doi":"https://doi.org/10.48550/arxiv.2604.25315","title":"SaliencyDecor: Enhancing Neural Network Interpretability through Feature Decorrelation","display_name":"SaliencyDecor: Enhancing Neural Network Interpretability through Feature Decorrelation","publication_year":2026,"publication_date":"2026-04-28","ids":{"openalex":"https://openalex.org/W7157604428","doi":"https://doi.org/10.48550/arxiv.2604.25315"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2604.25315","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.25315","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":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2604.25315","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5013688249","display_name":"Ali Karkehabadi","orcid":"https://orcid.org/0000-0001-6075-0880"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Karkehabadi, Ali","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5044138001","display_name":"Jamshid Hassanpour","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hassanpour, Jamshid","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134854101","display_name":"Houman Homayoun","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Homayoun, Houman","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5060036961","display_name":"Avesta Sasan","orcid":"https://orcid.org/0000-0002-4052-8075"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sasan, Avesta","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"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":0.9886000156402588,"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":0.9886000156402588,"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":0.002099999925121665,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.0020000000949949026,"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/interpretability","display_name":"Interpretability","score":0.9690999984741211},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.7235000133514404},{"id":"https://openalex.org/keywords/decorrelation","display_name":"Decorrelation","score":0.6488999724388123},{"id":"https://openalex.org/keywords/fidelity","display_name":"Fidelity","score":0.5723999738693237},{"id":"https://openalex.org/keywords/consistency","display_name":"Consistency (knowledge bases)","score":0.5360999703407288},{"id":"https://openalex.org/keywords/pattern-recognition","display_name":"Pattern recognition (psychology)","score":0.5338000059127808},{"id":"https://openalex.org/keywords/flexibility","display_name":"Flexibility (engineering)","score":0.4799000024795532},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.46160000562667847}],"concepts":[{"id":"https://openalex.org/C2781067378","wikidata":"https://www.wikidata.org/wiki/Q17027399","display_name":"Interpretability","level":2,"score":0.9690999984741211},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.729200005531311},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.7235000133514404},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7044000029563904},{"id":"https://openalex.org/C177860922","wikidata":"https://www.wikidata.org/wiki/Q788608","display_name":"Decorrelation","level":2,"score":0.6488999724388123},{"id":"https://openalex.org/C2776459999","wikidata":"https://www.wikidata.org/wiki/Q2119376","display_name":"Fidelity","level":2,"score":0.5723999738693237},{"id":"https://openalex.org/C2776436953","wikidata":"https://www.wikidata.org/wiki/Q5163215","display_name":"Consistency (knowledge bases)","level":2,"score":0.5360999703407288},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.5338000059127808},{"id":"https://openalex.org/C2780598303","wikidata":"https://www.wikidata.org/wiki/Q65921492","display_name":"Flexibility (engineering)","level":2,"score":0.4799000024795532},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.46309998631477356},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.46160000562667847},{"id":"https://openalex.org/C83665646","wikidata":"https://www.wikidata.org/wiki/Q42139305","display_name":"Feature vector","level":2,"score":0.39089998602867126},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.34299999475479126},{"id":"https://openalex.org/C148483581","wikidata":"https://www.wikidata.org/wiki/Q446488","display_name":"Feature selection","level":2,"score":0.3239000141620636},{"id":"https://openalex.org/C2984842247","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep neural networks","level":3,"score":0.30070000886917114},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.2948000133037567},{"id":"https://openalex.org/C59404180","wikidata":"https://www.wikidata.org/wiki/Q17013334","display_name":"Feature learning","level":2,"score":0.28760001063346863},{"id":"https://openalex.org/C52622490","wikidata":"https://www.wikidata.org/wiki/Q1026626","display_name":"Feature extraction","level":2,"score":0.2870999872684479},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.2791000008583069},{"id":"https://openalex.org/C2776135515","wikidata":"https://www.wikidata.org/wiki/Q17143721","display_name":"Regularization (linguistics)","level":2,"score":0.27140000462532043}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2604.25315","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.25315","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:10.48550/arxiv.2604.25315","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.25315","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":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Gradient-based":[0],"saliency":[1,57,90,153],"methods":[2,91],"are":[3],"widely":[4],"used":[5],"to":[6,84],"interpret":[7],"deep":[8],"neural":[9],"networks,":[10],"yet":[11],"they":[12],"often":[13],"produce":[14],"noisy":[15],"and":[16,55,74,109,125,141,150,179,188],"unstable":[17],"explanations":[18],"that":[19,29,80,144],"poorly":[20],"align":[21],"with":[22],"semantically":[23],"meaningful":[24],"input":[25],"features.":[26],"We":[27],"argue":[28],"a":[30,68,77,126,172],"fundamental":[31],"cause":[32],"of":[33,40,71,113],"this":[34,61],"behavior":[35],"lies":[36],"in":[37,53],"the":[38,97,111,164,182],"geometry":[39],"learned":[41],"representations:":[42],"correlated":[43],"feature":[44,65,82,98,123],"dimensions":[45],"diffuse":[46],"attribution":[47,86],"gradients":[48],"across":[49,138,163],"redundant":[50],"directions,":[51],"resulting":[52],"blurred":[54],"unreliable":[56],"maps.":[58],"To":[59],"address":[60],"issue,":[62],"we":[63],"identify":[64],"correlation":[66],"as":[67,171],"structural":[69],"limitation":[70],"gradient-based":[72],"interpretability":[73,178],"propose":[75],"SaliencyDecor,":[76],"training":[78],"framework":[79],"enforces":[81],"decorrelation":[83,127],"improve":[85],"fidelity":[87,112],"without":[88],"modifying":[89],"or":[92,133],"model":[93,189],"architectures":[94,142],"by":[95],"reshaping":[96],"space":[99],"toward":[100],"orthogonality,":[101],"our":[102,145,169],"approach":[103],"promotes":[104],"more":[105,151],"concentrated":[106],"gradient":[107],"flow":[108],"improves":[110],"saliency-based":[114],"explanations.":[115],"SaliencyDecor":[116],"jointly":[117],"optimizes":[118],"classification,":[119],"prediction":[120],"consistency":[121],"under":[122],"masking,":[124],"regularizer,":[128],"requiring":[129],"no":[130],"architectural":[131],"changes":[132],"inference-time":[134],"overhead.":[135],"Extensive":[136],"experiments":[137],"multiple":[139],"benchmarks":[140],"demonstrate":[143],"method":[146,170],"produces":[147],"substantially":[148],"sharper":[149],"object-focused":[152],"maps":[154],"while":[155],"simultaneously":[156],"improving":[157],"predictive":[158],"performance,":[159],"achieving":[160],"accuracy":[161],"gains":[162],"datasets.":[165],"These":[166],"results":[167],"establish":[168],"principled":[173],"mechanism":[174],"for":[175],"enhancing":[176],"both":[177],"accuracy,":[180],"challenging":[181],"conventional":[183],"trade-off":[184],"between":[185],"explanation":[186],"quality":[187],"performance.":[190]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-04-30T00:00:00"}