{"id":"https://openalex.org/W7159602978","doi":"https://doi.org/10.48550/arxiv.2604.27870","title":"Parameter-Efficient Architectural Modifications for Translation-Invariant CNNs","display_name":"Parameter-Efficient Architectural Modifications for Translation-Invariant CNNs","publication_year":2026,"publication_date":"2026-04-30","ids":{"openalex":"https://openalex.org/W7159602978","doi":"https://doi.org/10.48550/arxiv.2604.27870"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2604.27870","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.27870","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"article"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2604.27870","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5107375803","display_name":"Nuria Alabau-Bosque","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Alabau-Bosque, Nuria","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5032878706","display_name":"Jorge Vila-Tom\u00e1s","orcid":"https://orcid.org/0000-0002-8246-0195"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Vila-Tomas, Jorge","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5107524466","display_name":"Paula Daud\u00e9n-Oliver","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Dauden-Oliver, Paula","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5065281659","display_name":"Valero Laparra","orcid":"https://orcid.org/0000-0001-7531-9890"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Laparra, Valero","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5134986070","display_name":"Jesus Malo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Malo, Jesus","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":5,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.6359000205993652,"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"}},"topics":[{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.6359000205993652,"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":0.17790000140666962,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.02850000001490116,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/pooling","display_name":"Pooling","score":0.6851000189781189},{"id":"https://openalex.org/keywords/robustness","display_name":"Robustness (evolution)","score":0.6108999848365784},{"id":"https://openalex.org/keywords/convolutional-neural-network","display_name":"Convolutional neural network","score":0.5620999932289124},{"id":"https://openalex.org/keywords/pattern-recognition","display_name":"Pattern recognition (psychology)","score":0.4787999987602234},{"id":"https://openalex.org/keywords/residual","display_name":"Residual","score":0.45159998536109924},{"id":"https://openalex.org/keywords/reduction","display_name":"Reduction (mathematics)","score":0.4422999918460846},{"id":"https://openalex.org/keywords/metric","display_name":"Metric (unit)","score":0.42750000953674316},{"id":"https://openalex.org/keywords/invariant","display_name":"Invariant (physics)","score":0.41659998893737793},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.41519999504089355}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7529000043869019},{"id":"https://openalex.org/C70437156","wikidata":"https://www.wikidata.org/wiki/Q7228652","display_name":"Pooling","level":2,"score":0.6851000189781189},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.6108999848365784},{"id":"https://openalex.org/C81363708","wikidata":"https://www.wikidata.org/wiki/Q17084460","display_name":"Convolutional neural network","level":2,"score":0.5620999932289124},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.526199996471405},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.4787999987602234},{"id":"https://openalex.org/C155512373","wikidata":"https://www.wikidata.org/wiki/Q287450","display_name":"Residual","level":2,"score":0.45159998536109924},{"id":"https://openalex.org/C111335779","wikidata":"https://www.wikidata.org/wiki/Q3454686","display_name":"Reduction (mathematics)","level":2,"score":0.4422999918460846},{"id":"https://openalex.org/C176217482","wikidata":"https://www.wikidata.org/wiki/Q860554","display_name":"Metric (unit)","level":2,"score":0.42750000953674316},{"id":"https://openalex.org/C190470478","wikidata":"https://www.wikidata.org/wiki/Q2370229","display_name":"Invariant (physics)","level":2,"score":0.41659998893737793},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.41519999504089355},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3986999988555908},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.39800000190734863},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3887999951839447},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.3181999921798706},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.3160000145435333},{"id":"https://openalex.org/C193415008","wikidata":"https://www.wikidata.org/wiki/Q639681","display_name":"Network architecture","level":2,"score":0.3052999973297119},{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.3052000105381012},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.3043000102043152},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.3025999963283539},{"id":"https://openalex.org/C3826847","wikidata":"https://www.wikidata.org/wiki/Q188768","display_name":"FLOPS","level":2,"score":0.30090001225471497},{"id":"https://openalex.org/C151201525","wikidata":"https://www.wikidata.org/wiki/Q177239","display_name":"Limit (mathematics)","level":2,"score":0.2912999987602234},{"id":"https://openalex.org/C89600930","wikidata":"https://www.wikidata.org/wiki/Q1423946","display_name":"Segmentation","level":2,"score":0.28700000047683716},{"id":"https://openalex.org/C95713431","wikidata":"https://www.wikidata.org/wiki/Q631425","display_name":"Vulnerability (computing)","level":2,"score":0.28529998660087585},{"id":"https://openalex.org/C110384440","wikidata":"https://www.wikidata.org/wiki/Q1143270","display_name":"Upsampling","level":3,"score":0.28189998865127563},{"id":"https://openalex.org/C2776036281","wikidata":"https://www.wikidata.org/wiki/Q48769818","display_name":"Constraint (computer-aided design)","level":2,"score":0.26269999146461487},{"id":"https://openalex.org/C2780898871","wikidata":"https://www.wikidata.org/wiki/Q860554","display_name":"Performance metric","level":2,"score":0.2572000026702881}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2604.27870","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.27870","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"article"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2604.27870","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.27870","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Convolutional":[0],"Neural":[1],"Networks":[2],"(CNNs)":[3],"are":[4,237,244],"widely":[5],"assumed":[6],"to":[7,26,92,104,129,138,163,228,247],"be":[8],"translation-invariant,":[9],"yet":[10],"standard":[11],"architectures":[12],"exhibit":[13],"a":[14,18,44,72,83,96,135,149,198,220],"startling":[15],"fragility:":[16],"even":[17],"single-pixel":[19],"shift":[20],"can":[21],"drastically":[22],"degrade":[23],"performance":[24],"due":[25],"their":[27],"reliance":[28],"on":[29,116,206],"spatially":[30],"dependent":[31],"fully":[32],"connected":[33],"layers.":[34],"In":[35],"this":[36,40,79,107],"work,":[37],"we":[38,61,76,159],"resolve":[39],"vulnerability":[41],"by":[42,169],"proposing":[43],"lightweight":[45],"'Online":[46],"Architecture'":[47],"strategy.":[48],"By":[49],"strategically":[50],"inserting":[51],"Global":[52],"Average":[53],"Pooling":[54],"(GAP)":[55],"layers":[56],"at":[57],"various":[58],"network":[59,101],"depths,":[60],"effectively":[62],"decouple":[63],"feature":[64],"recognition":[65],"from":[66,127],"spatial":[67],"location.":[68],"Using":[69],"VGG-16":[70],"as":[71],"primary":[73],"case":[74],"study,":[75],"demonstrate":[77],"that":[78,153,215],"architectural":[80,217],"modification":[81],"achieves":[82,197],"massive":[84],"98%":[85],"reduction":[86,98],"in":[87,99,186],"trainable":[88],"parameters":[89],"(from":[90],"5.2M":[91],"just":[93],"82K)":[94],"and":[95,196,224,235,242,250],"90%":[97],"total":[100],"size":[102],"(138M":[103],"14M).":[105],"Despite":[106],"drastic":[108],"pruning,":[109],"our":[110,132,171],"variants":[111],"maintain":[112],"competitive":[113],"Top-1":[114],"accuracy":[115],"ImageNet":[117],"(66.4%)":[118],"while":[119,140],"doubling":[120],"translational":[121],"robustness,":[122],"reducing":[123],"average":[124],"relative":[125],"loss":[126],"0.09":[128],"0.05.":[130],"Furthermore,":[131],"analysis":[133],"identifies":[134],"fundamental":[136],"limit":[137],"invariance:":[139],"GAP":[141],"resolves":[142],"macroscopic":[143],"sensitivity,":[144],"discrete":[145],"pooling":[146],"operations":[147],"introduce":[148],"residual":[150],"periodic":[151],"aliasing":[152],"prevents":[154],"perfect":[155],"pixel-level":[156],"stability.":[157],"Finally,":[158],"extend":[160],"these":[161],"findings":[162],"Perceptual":[164],"Image":[165],"Quality":[166],"Assessment":[167],"(IQA)":[168],"integrating":[170],"invariant":[172],"backbones":[173],"into":[174],"the":[175,183,189,207],"LPIPS":[176],"framework.":[177],"The":[178,240],"resulting":[179],"metric":[180],"significantly":[181],"outperforms":[182],"retrained":[184],"baseline":[185],"generalization":[187],"across":[188],"KADID-10k":[190],"dataset":[191,209],"(Spearman":[192,210],"0.89":[193],"vs.":[194],"0.75)":[195],"near-perfect":[199],"alignment":[200],"with":[201],"human":[202],"psychophysical":[203],"response":[204],"curves":[205],"RAID":[208],"0.95).":[211],"These":[212],"results":[213],"confirm":[214],"enforcing":[216],"invariance":[218],"is":[219],"far":[221],"more":[222],"efficient":[223],"biologically":[225],"plausible":[226],"path":[227],"robustness":[229],"than":[230],"traditional":[231],"data":[232,241],"augmentation.":[233],"Data":[234],"code":[236,243],"publicly":[238,245],"available.":[239],"available":[246],"facilitate":[248],"validation":[249],"further":[251],"research.":[252]},"counts_by_year":[],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2026-05-02T00:00:00"}