{"id":"https://openalex.org/W7164211082","doi":"https://doi.org/10.1109/fccm68464.2026.00019","title":"GraphLeap: Decoupling Graph Construction and Convolution for Vision GNN Acceleration on FPGA","display_name":"GraphLeap: Decoupling Graph Construction and Convolution for Vision GNN Acceleration on FPGA","publication_year":2026,"publication_date":"2026-05-13","ids":{"openalex":"https://openalex.org/W7164211082","doi":"https://doi.org/10.1109/fccm68464.2026.00019"},"language":null,"primary_location":{"id":"doi:10.1109/fccm68464.2026.00019","is_oa":false,"landing_page_url":"https://doi.org/10.1109/fccm68464.2026.00019","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2026 IEEE 34th Annual International Symposium on Field-Programmable Custom Computing Machines (FCCM)","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5134516845","display_name":"Anvitha Ramachandran","orcid":null},"institutions":[{"id":"https://openalex.org/I1174212","display_name":"University of Southern California","ror":"https://ror.org/03taz7m60","country_code":"US","type":"education","lineage":["https://openalex.org/I1174212"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Anvitha Ramachandran","raw_affiliation_strings":["University of Southern California,Los Angeles,California,USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Southern California,Los Angeles,California,USA","institution_ids":["https://openalex.org/I1174212"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134529200","display_name":"Dhruv Parikh","orcid":null},"institutions":[{"id":"https://openalex.org/I1174212","display_name":"University of Southern California","ror":"https://ror.org/03taz7m60","country_code":"US","type":"education","lineage":["https://openalex.org/I1174212"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Dhruv Parikh","raw_affiliation_strings":["University of Southern California,Los Angeles,California,USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Southern California,Los Angeles,California,USA","institution_ids":["https://openalex.org/I1174212"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5138337899","display_name":"Viktor Prasanna","orcid":null},"institutions":[{"id":"https://openalex.org/I1174212","display_name":"University of Southern California","ror":"https://ror.org/03taz7m60","country_code":"US","type":"education","lineage":["https://openalex.org/I1174212"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Viktor Prasanna","raw_affiliation_strings":["University of Southern California,Los Angeles,California,USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Southern California,Los Angeles,California,USA","institution_ids":["https://openalex.org/I1174212"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I1174212"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.85203409,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"38","last_page":"47"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12292","display_name":"Graph Theory and Algorithms","score":0.8123000264167786,"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/T12292","display_name":"Graph Theory and Algorithms","score":0.8123000264167786,"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/T11273","display_name":"Advanced Graph Neural Networks","score":0.06650000065565109,"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":0.017500000074505806,"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/decoupling","display_name":"Decoupling (probability)","score":0.5442000031471252},{"id":"https://openalex.org/keywords/graph","display_name":"Graph","score":0.49779999256134033},{"id":"https://openalex.org/keywords/acceleration","display_name":"Acceleration","score":0.4794999957084656},{"id":"https://openalex.org/keywords/convolution","display_name":"Convolution (computer science)","score":0.47380000352859497}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6351000070571899},{"id":"https://openalex.org/C205606062","wikidata":"https://www.wikidata.org/wiki/Q5249645","display_name":"Decoupling (probability)","level":2,"score":0.5442000031471252},{"id":"https://openalex.org/C132525143","wikidata":"https://www.wikidata.org/wiki/Q141488","display_name":"Graph","level":2,"score":0.49779999256134033},{"id":"https://openalex.org/C117896860","wikidata":"https://www.wikidata.org/wiki/Q11376","display_name":"Acceleration","level":2,"score":0.4794999957084656},{"id":"https://openalex.org/C45347329","wikidata":"https://www.wikidata.org/wiki/Q5166604","display_name":"Convolution (computer science)","level":3,"score":0.47380000352859497},{"id":"https://openalex.org/C42935608","wikidata":"https://www.wikidata.org/wiki/Q190411","display_name":"Field-programmable gate array","level":2,"score":0.44670000672340393},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.4275999963283539},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.38960000872612},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.3783000111579895},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.2985999882221222}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/fccm68464.2026.00019","is_oa":false,"landing_page_url":"https://doi.org/10.1109/fccm68464.2026.00019","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2026 IEEE 34th Annual International Symposium on Field-Programmable Custom Computing Machines (FCCM)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":28,"referenced_works":["https://openalex.org/W1501856433","https://openalex.org/W2108598243","https://openalex.org/W2116341502","https://openalex.org/W2990045899","https://openalex.org/W3017228913","https://openalex.org/W3176940011","https://openalex.org/W4200197440","https://openalex.org/W4283796372","https://openalex.org/W4327503237","https://openalex.org/W4381327585","https://openalex.org/W4385815503","https://openalex.org/W4385960790","https://openalex.org/W4386568606","https://openalex.org/W4386764001","https://openalex.org/W4390873987","https://openalex.org/W4393407124","https://openalex.org/W4402349219","https://openalex.org/W4402754254","https://openalex.org/W4402903991","https://openalex.org/W4404455164","https://openalex.org/W4405718104","https://openalex.org/W4406893597","https://openalex.org/W4408222700","https://openalex.org/W4409766690","https://openalex.org/W4411725841","https://openalex.org/W4413147804","https://openalex.org/W4413393364","https://openalex.org/W4415250774"],"related_works":[],"abstract_inverted_index":{"Vision":[0,25,211,281],"Graph":[1],"Neural":[2],"Networks":[3],"(ViGs)":[4],"model":[5],"an":[6,257],"image":[7],"as":[8],"a":[9,40,86,100,127,164,191,216,222,228],"graph":[10,35,42,60,68,90,111,128,145,153,157,178,224],"of":[11,67,83,167,245,279],"patch":[12,52],"tokens,":[13],"enabling":[14],"adaptive,":[15],"feature-driven":[16],"neighborhoods.":[17],"Unlike":[18],"CNNs":[19],"with":[20,27,227],"fixed":[21],"grid":[22],"biases":[23],"or":[24],"Transformers":[26],"global":[28],"token":[29],"interactions,":[30],"ViGs":[31],"rely":[32],"on":[33,50,71,201,249,256],"dynamic":[34,152],"convolution:":[36],"at":[37,123,297],"each":[38],"layer,":[39],"feature-dependent":[41],"is":[43,62,80,194,215,295],"built":[44],"via":[45],"k-nearest":[46],"neighbor":[47],"(kNN)":[48],"search":[49],"current":[51,139],"features,":[53,134],"followed":[54],"by":[55,76,109],"message":[56,161],"passing.":[57],"This":[58,149],"per-layer":[59],"construction":[61,91,112,154,158,179,225],"the":[63,81,120,131,138,144,205,277],"main":[64],"bottleneck\u2014consuming":[65],"50\u201395%":[66],"convolution":[69],"time":[70],"CPUs":[72],"and":[73,92,159,234,237,251,270,284],"GPUs\u2014and":[74],"scales":[75],"O(N2)":[77],"where":[78],"N":[79],"number":[82],"patches,":[84],"creating":[85],"sequential":[87],"dependency":[88,108],"between":[89],"feature":[93,114,121,229],"updates.In":[94],"this":[95,107],"paper,":[96],"we":[97,170,185,203],"introduce":[98,181],"GraphLeap,":[99,202],"simple":[101],"yet":[102],"novel":[103],"reformulation":[104],"that":[105,187,220],"removes":[106],"decoupling":[110],"from":[113,130],"update":[115,122,230],"across":[116],"layers.":[117],"GraphLeap":[118],"performs":[119],"layer":[124,147],"\u2113":[125],"using":[126,137,174],"constructed":[129],"previous":[132],"layer\u2019s":[133,140],"while":[135],"simultaneously":[136],"features":[141,176],"to":[142,196,265],"construct":[143],"for":[146,177,190,210,290],"\u2113+1.":[148],"one-layer":[150],"lookahead":[151],"enables":[155,238],"concurrent":[156],"GNN":[160,282],"passing,":[162],"yielding":[163],"new":[165],"class":[166],"ViGs,":[168],"which":[169],"call":[171],"GraphLeap.":[172],"While":[173],"prior-layer":[175],"can":[180],"minor":[182],"accuracy":[183],"degradation,":[184],"show":[186],"lightweight":[188],"fine-tuning":[189],"few":[192],"epochs":[193],"sufficient":[195],"recover":[197],"original":[198],"accuracy.":[199],"Building":[200],"present":[204],"first":[206],"end-to-end":[207],"FPGA":[208],"accelerator":[209],"GNNs.":[212],"Our":[213],"design":[214],"streaming,":[217],"layer-pipelined":[218],"architecture":[219],"overlaps":[221],"kNN":[223],"engine":[226],"engine,":[231],"exploits":[232],"node":[233],"channel-level":[235],"parallelism,":[236],"efficient":[239],"on-chip":[240],"dataflow":[241],"without":[242],"explicit":[243],"materialization":[244],"edge":[246],"features.":[247],"Evaluated":[248],"isotropic":[250],"pyramidal":[252],"ViG":[253],"models":[254],"deployed":[255],"Alveo":[258],"U280":[259],"FPGA,":[260],"our":[261],"approach":[262],"achieves":[263],"up":[264],"95.7\u00d7":[266],"speedup":[267,272],"over":[268,273],"CPU":[269],"8.5\u00d7":[271],"GPU":[274],"baselines,":[275],"demonstrating":[276],"feasibility":[278],"real-time":[280],"inference":[283],"motivating":[285],"future":[286],"hardware\u2013":[287],"algorithm":[288],"co-design":[289],"graph-based":[291],"vision":[292],"models.":[293],"Code":[294],"available":[296],"https://github.com/anvitha305/GraphLeap.":[298]},"counts_by_year":[],"updated_date":"2026-06-26T08:34:08.712188","created_date":"2026-06-11T00:00:00"}