{"id":"https://openalex.org/W4404133642","doi":"https://doi.org/10.1145/3649329.3657328","title":"DEFA: Efficient Deformable Attention Acceleration via Pruning-Assisted Grid-Sampling and Multi-Scale Parallel Processing","display_name":"DEFA: Efficient Deformable Attention Acceleration via Pruning-Assisted Grid-Sampling and Multi-Scale Parallel Processing","publication_year":2024,"publication_date":"2024-06-23","ids":{"openalex":"https://openalex.org/W4404133642","doi":"https://doi.org/10.1145/3649329.3657328"},"language":"en","primary_location":{"id":"doi:10.1145/3649329.3657328","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3649329.3657328","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 61st ACM/IEEE Design Automation Conference","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/A5074812206","display_name":"Yansong Xu","orcid":"https://orcid.org/0009-0008-4511-7436"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":true,"raw_author_name":"Yansong Xu","raw_affiliation_strings":["Shanghai Jiao Tong University, Shanghai, Shanghai, China"],"raw_orcid":"https://orcid.org/0009-0008-4511-7436","affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University, Shanghai, Shanghai, China","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5066692868","display_name":"Dongxu Lyu","orcid":"https://orcid.org/0000-0001-6826-2670"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Dongxu Lyu","raw_affiliation_strings":["Shanghai Jiao Tong University, Shanghai, Shanghai, China"],"raw_orcid":"https://orcid.org/0000-0001-6826-2670","affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University, Shanghai, Shanghai, China","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5032381238","display_name":"Zhenyu Li","orcid":"https://orcid.org/0009-0006-9212-6904"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhenyu Li","raw_affiliation_strings":["Shanghai Jiao Tong University, Shanghai, Shanghai, China"],"raw_orcid":"https://orcid.org/0009-0006-9212-6904","affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University, Shanghai, Shanghai, China","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5010349853","display_name":"Yuzhou Chen","orcid":"https://orcid.org/0009-0004-9236-0480"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yuzhou Chen","raw_affiliation_strings":["Shanghai Jiao Tong University, Shanghai, Shanghai, China"],"raw_orcid":"https://orcid.org/0009-0004-9236-0480","affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University, Shanghai, Shanghai, China","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5115602320","display_name":"Zilong Wang","orcid":"https://orcid.org/0009-0001-8882-2345"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zilong Wang","raw_affiliation_strings":["Shanghai Jiao Tong University, Shanghai, Shanghai, China"],"raw_orcid":"https://orcid.org/0009-0001-8882-2345","affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University, Shanghai, Shanghai, China","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5006070507","display_name":"Gang Wang","orcid":"https://orcid.org/0009-0003-6944-2958"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Gang Wang","raw_affiliation_strings":["Shanghai Jiao Tong University, Shanghai, Shanghai, China"],"raw_orcid":"https://orcid.org/0009-0003-6944-2958","affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University, Shanghai, Shanghai, China","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5060603700","display_name":"Zhican Wang","orcid":"https://orcid.org/0000-0001-6051-9486"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhican Wang","raw_affiliation_strings":["Shanghai Jiao Tong University, Shanghai, Shanghai, China"],"raw_orcid":"https://orcid.org/0000-0001-6051-9486","affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University, Shanghai, Shanghai, China","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101399528","display_name":"Haoming Li","orcid":"https://orcid.org/0000-0002-2939-6534"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Haomin Li","raw_affiliation_strings":["Shanghai Jiao Tong University, Shanghai, Shanghai, China"],"raw_orcid":"https://orcid.org/0000-0002-2939-6534","affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University, Shanghai, Shanghai, China","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5054897331","display_name":"Guanghui He","orcid":"https://orcid.org/0000-0002-0486-6421"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Guanghui He","raw_affiliation_strings":["Shanghai Jiao Tong University, Shanghai, Shanghai, China"],"raw_orcid":"https://orcid.org/0000-0002-0486-6421","affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University, Shanghai, Shanghai, China","institution_ids":["https://openalex.org/I183067930"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":9,"corresponding_author_ids":["https://openalex.org/A5074812206"],"corresponding_institution_ids":["https://openalex.org/I183067930"],"apc_list":null,"apc_paid":null,"fwci":0.9523,"has_fulltext":false,"cited_by_count":4,"citation_normalized_percentile":{"value":0.77196045,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":97,"max":98},"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"6"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9993000030517578,"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":0.9993000030517578,"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/T11992","display_name":"CCD and CMOS Imaging Sensors","score":0.9991999864578247,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10502","display_name":"Advanced Memory and Neural Computing","score":0.9973999857902527,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7867303490638733},{"id":"https://openalex.org/keywords/acceleration","display_name":"Acceleration","score":0.7167690992355347},{"id":"https://openalex.org/keywords/pruning","display_name":"Pruning","score":0.698377251625061},{"id":"https://openalex.org/keywords/grid","display_name":"Grid","score":0.6158376336097717},{"id":"https://openalex.org/keywords/scale","display_name":"Scale (ratio)","score":0.5162441730499268},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.510831356048584},{"id":"https://openalex.org/keywords/sampling","display_name":"Sampling (signal processing)","score":0.4686547815799713},{"id":"https://openalex.org/keywords/parallel-processing","display_name":"Parallel processing","score":0.42100125551223755},{"id":"https://openalex.org/keywords/computational-science","display_name":"Computational science","score":0.40048763155937195},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.3349844813346863},{"id":"https://openalex.org/keywords/computer-vision","display_name":"Computer vision","score":0.198146253824234},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.07731184363365173},{"id":"https://openalex.org/keywords/cartography","display_name":"Cartography","score":0.05096304416656494},{"id":"https://openalex.org/keywords/geography","display_name":"Geography","score":0.04740700125694275},{"id":"https://openalex.org/keywords/geometry","display_name":"Geometry","score":0.04554054141044617},{"id":"https://openalex.org/keywords/physics","display_name":"Physics","score":0.04480084776878357}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7867303490638733},{"id":"https://openalex.org/C117896860","wikidata":"https://www.wikidata.org/wiki/Q11376","display_name":"Acceleration","level":2,"score":0.7167690992355347},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.698377251625061},{"id":"https://openalex.org/C187691185","wikidata":"https://www.wikidata.org/wiki/Q2020720","display_name":"Grid","level":2,"score":0.6158376336097717},{"id":"https://openalex.org/C2778755073","wikidata":"https://www.wikidata.org/wiki/Q10858537","display_name":"Scale (ratio)","level":2,"score":0.5162441730499268},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.510831356048584},{"id":"https://openalex.org/C140779682","wikidata":"https://www.wikidata.org/wiki/Q210868","display_name":"Sampling (signal processing)","level":3,"score":0.4686547815799713},{"id":"https://openalex.org/C106515295","wikidata":"https://www.wikidata.org/wiki/Q26806595","display_name":"Parallel processing","level":2,"score":0.42100125551223755},{"id":"https://openalex.org/C459310","wikidata":"https://www.wikidata.org/wiki/Q117801","display_name":"Computational science","level":1,"score":0.40048763155937195},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3349844813346863},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.198146253824234},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.07731184363365173},{"id":"https://openalex.org/C58640448","wikidata":"https://www.wikidata.org/wiki/Q42515","display_name":"Cartography","level":1,"score":0.05096304416656494},{"id":"https://openalex.org/C205649164","wikidata":"https://www.wikidata.org/wiki/Q1071","display_name":"Geography","level":0,"score":0.04740700125694275},{"id":"https://openalex.org/C2524010","wikidata":"https://www.wikidata.org/wiki/Q8087","display_name":"Geometry","level":1,"score":0.04554054141044617},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.04480084776878357},{"id":"https://openalex.org/C74650414","wikidata":"https://www.wikidata.org/wiki/Q11397","display_name":"Classical mechanics","level":1,"score":0.0},{"id":"https://openalex.org/C6557445","wikidata":"https://www.wikidata.org/wiki/Q173113","display_name":"Agronomy","level":1,"score":0.0},{"id":"https://openalex.org/C86803240","wikidata":"https://www.wikidata.org/wiki/Q420","display_name":"Biology","level":0,"score":0.0},{"id":"https://openalex.org/C106131492","wikidata":"https://www.wikidata.org/wiki/Q3072260","display_name":"Filter (signal processing)","level":2,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3649329.3657328","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3649329.3657328","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 61st ACM/IEEE Design Automation Conference","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"display_name":"Affordable and clean energy","id":"https://metadata.un.org/sdg/7","score":0.7099999785423279}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":12,"referenced_works":["https://openalex.org/W1861492603","https://openalex.org/W2072730350","https://openalex.org/W2601564443","https://openalex.org/W3091813304","https://openalex.org/W3096609285","https://openalex.org/W3132460984","https://openalex.org/W3159727696","https://openalex.org/W4308067493","https://openalex.org/W4312894406","https://openalex.org/W4313467238","https://openalex.org/W4386066358","https://openalex.org/W4386083035"],"related_works":["https://openalex.org/W2373300491","https://openalex.org/W2395294869","https://openalex.org/W2378744544","https://openalex.org/W2565094479","https://openalex.org/W2594301978","https://openalex.org/W2390829436","https://openalex.org/W2379704676","https://openalex.org/W1998810860","https://openalex.org/W4206442282","https://openalex.org/W1989791859"],"abstract_inverted_index":{"Multi-scale":[0],"deformable":[1],"attention":[2,43],"(MSDeformAttn)":[3],"has":[4],"emerged":[5],"as":[6],"a":[7,63],"key":[8],"mechanism":[9],"in":[10],"various":[11],"vision":[12],"tasks,":[13],"demonstrating":[14],"explicit":[15],"superiority":[16],"attributed":[17],"to":[18,35,49,52,109,143],"multi-scale":[19,107],"grid-sampling.":[20],"However,":[21],"this":[22,57],"newly":[23],"introduced":[24],"operator":[25],"incurs":[26],"irregular":[27],"data":[28],"access":[29,119],"and":[30,83,89,114,124,137],"enormous":[31],"memory":[32,95,118],"requirement,":[33],"leading":[34],"severe":[36],"PE":[37],"under-utilization.":[38],"Meanwhile,":[39],"existing":[40],"approaches":[41],"for":[42,56,72,86],"acceleration":[44],"cannot":[45],"be":[46],"directly":[47],"applied":[48],"MSDeformAttn":[50,73],"due":[51],"lack":[53],"of":[54,162],"support":[55,161],"distinct":[58],"procedure.":[59],"Therefore,":[60],"we":[61],"propose":[62],"dedicated":[64],"algorithm-architecture":[65],"co-design":[66],"dubbed":[67],"DEFA,":[68],"the":[69,76,94,101,106,111,117,150],"first-of-its-kind":[70],"method":[71],"acceleration.":[74],"At":[75,100],"algorithm":[77],"level,":[78,103],"DEFA":[79,133],"adopts":[80],"frequency-weighted":[81],"pruning":[82,85],"probability-aware":[84],"feature":[87,125],"maps":[88],"sampling":[90],"points":[91],"respectively,":[92],"alleviating":[93],"footprint":[96],"by":[97,153],"over":[98],"80%.":[99],"architecture":[102],"it":[104],"explores":[105],"parallelism":[108],"boost":[110,141],"throughput":[112],"significantly":[113],"further":[115],"reduces":[116],"via":[120],"fine-grained":[121],"layer":[122],"fusion":[123],"map":[126],"reusing.":[127],"Extensively":[128],"evaluated":[129],"on":[130],"representative":[131],"benchmarks,":[132],"achieves":[134],"10.1-31.9X":[135],"speedup":[136],"20.3-37.7X":[138],"energy":[139,155],"efficiency":[140,156],"compared":[142],"powerful":[144],"GPU":[145],"platforms.":[146],"It":[147],"also":[148],"rivals":[149],"related":[151],"accelerators":[152],"2.2-3.7X":[154],"improvement":[157],"while":[158],"providing":[159],"pioneering":[160],"MSDeformAttn.":[163]},"counts_by_year":[{"year":2025,"cited_by_count":4}],"updated_date":"2025-12-21T01:58:51.020947","created_date":"2025-10-10T00:00:00"}