HSBNN: A High-Scalable Bayesian Neural Networks Accelerator Based on Field Programmable Gate Arrays (FPGA)
Traditional artificial neural networks have inherent overfitting problems and tend to produce overly confident predictions due to their reliance on point estimation methods. In contrast, Bayesian theory offers a probabilistic framework that replaces point estimation with probability distributions, e...
Uložené v:
| Vydané v: | Cognitive computation Ročník 17; číslo 3; s. 100 |
|---|---|
| Hlavní autori: | , , , , , , , |
| Médium: | Journal Article |
| Jazyk: | English |
| Vydavateľské údaje: |
New York
Springer US
01.06.2025
Springer Nature B.V |
| Predmet: | |
| ISSN: | 1866-9956, 1866-9964 |
| On-line prístup: | Získať plný text |
| Tagy: |
Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
|
| Abstract | Traditional artificial neural networks have inherent overfitting problems and tend to produce overly confident predictions due to their reliance on point estimation methods. In contrast, Bayesian theory offers a probabilistic framework that replaces point estimation with probability distributions, effectively addressing issues of overconfidence. The brain is also believed working under the Bayesian rules, the neural networks of which evaluate the precision of prior knowledge and incoming evidence, achieving the balance of weight updating to the most reliable information sources [
1
]. By integrating Bayesian principles with artificial neural networks, the bio-inspired Bayesian Neural Networks (BNNs) can generate predictions accompanied by confidence evaluations, enhancing their practical applicability. To further improve the computational efficiency of BNNs and enable scalable deployment on edge devices, we propose a High-Scalable Bayesian Neural Network (HSBNN) accelerator based on field-programmable gate arrays (FPGAs) with multiple optimizations. A resource-saving Gaussian random number generator (RS-GRNG) optimized for FPGAs shows high efficiency, which seamlessly extends to support parallel sampling of weight distributions, enabling reliable confidence probability evaluations. Furthermore, the parameterization of BNN architectures with configuration files and employment of a layer-by-layer computing mode ensure that different BNNs can be accelerated without reprogramming the FPGA, offering excellent scalability. The entire system, implemented with the OpenCL heterogeneous computing library, leverages parallel processing units and pipeline channels to achieve high acceleration performance and efficient data transfer. The experiment results demonstrate that the system achieves a data processing throughput of 1.002 milliseconds per image, exceeding CPU performance by 1000-fold and GPU performance by nearly 500-fold. |
|---|---|
| AbstractList | Traditional artificial neural networks have inherent overfitting problems and tend to produce overly confident predictions due to their reliance on point estimation methods. In contrast, Bayesian theory offers a probabilistic framework that replaces point estimation with probability distributions, effectively addressing issues of overconfidence. The brain is also believed working under the Bayesian rules, the neural networks of which evaluate the precision of prior knowledge and incoming evidence, achieving the balance of weight updating to the most reliable information sources [1]. By integrating Bayesian principles with artificial neural networks, the bio-inspired Bayesian Neural Networks (BNNs) can generate predictions accompanied by confidence evaluations, enhancing their practical applicability. To further improve the computational efficiency of BNNs and enable scalable deployment on edge devices, we propose a High-Scalable Bayesian Neural Network (HSBNN) accelerator based on field-programmable gate arrays (FPGAs) with multiple optimizations. A resource-saving Gaussian random number generator (RS-GRNG) optimized for FPGAs shows high efficiency, which seamlessly extends to support parallel sampling of weight distributions, enabling reliable confidence probability evaluations. Furthermore, the parameterization of BNN architectures with configuration files and employment of a layer-by-layer computing mode ensure that different BNNs can be accelerated without reprogramming the FPGA, offering excellent scalability. The entire system, implemented with the OpenCL heterogeneous computing library, leverages parallel processing units and pipeline channels to achieve high acceleration performance and efficient data transfer. The experiment results demonstrate that the system achieves a data processing throughput of 1.002 milliseconds per image, exceeding CPU performance by 1000-fold and GPU performance by nearly 500-fold. Traditional artificial neural networks have inherent overfitting problems and tend to produce overly confident predictions due to their reliance on point estimation methods. In contrast, Bayesian theory offers a probabilistic framework that replaces point estimation with probability distributions, effectively addressing issues of overconfidence. The brain is also believed working under the Bayesian rules, the neural networks of which evaluate the precision of prior knowledge and incoming evidence, achieving the balance of weight updating to the most reliable information sources [ 1 ]. By integrating Bayesian principles with artificial neural networks, the bio-inspired Bayesian Neural Networks (BNNs) can generate predictions accompanied by confidence evaluations, enhancing their practical applicability. To further improve the computational efficiency of BNNs and enable scalable deployment on edge devices, we propose a High-Scalable Bayesian Neural Network (HSBNN) accelerator based on field-programmable gate arrays (FPGAs) with multiple optimizations. A resource-saving Gaussian random number generator (RS-GRNG) optimized for FPGAs shows high efficiency, which seamlessly extends to support parallel sampling of weight distributions, enabling reliable confidence probability evaluations. Furthermore, the parameterization of BNN architectures with configuration files and employment of a layer-by-layer computing mode ensure that different BNNs can be accelerated without reprogramming the FPGA, offering excellent scalability. The entire system, implemented with the OpenCL heterogeneous computing library, leverages parallel processing units and pipeline channels to achieve high acceleration performance and efficient data transfer. The experiment results demonstrate that the system achieves a data processing throughput of 1.002 milliseconds per image, exceeding CPU performance by 1000-fold and GPU performance by nearly 500-fold. |
| ArticleNumber | 100 |
| Author | Lu, Wenyi Solé-Casals, Jordi Liu, Yinghao Ma, Yuan Zhang, Hao Duan, Feng Sun, Zhe Caiafa, Cesar F. |
| Author_xml | – sequence: 1 givenname: Yinghao surname: Liu fullname: Liu, Yinghao organization: Graduate School of Arts and Sciences, The University of Tokyo – sequence: 2 givenname: Hao surname: Zhang fullname: Zhang, Hao email: zhangh@szlab.ac.cn organization: Gusu Laboratory of Materials – sequence: 3 givenname: Zhe orcidid: 0000-0002-6531-0769 surname: Sun fullname: Sun, Zhe email: z.sun.kc@juntendo.ac.jp organization: Faculty of Health Data Science, Juntendo University – sequence: 4 givenname: Feng surname: Duan fullname: Duan, Feng email: duanf@nankai.edu.cn organization: College of Artificial Intelligence, Nankai University – sequence: 5 givenname: Yuan surname: Ma fullname: Ma, Yuan organization: Institute of Information Engineering, Chinese Academy of Sciences – sequence: 6 givenname: Wenyi surname: Lu fullname: Lu, Wenyi organization: College of Artificial Intelligence, Nankai University – sequence: 7 givenname: Cesar F. surname: Caiafa fullname: Caiafa, Cesar F. organization: Instituto Argentino de Radioastronomía, CONICET CCT La Plata/CIC-PBA/UNLP – sequence: 8 givenname: Jordi surname: Solé-Casals fullname: Solé-Casals, Jordi organization: Data and Signal Processing Research Group, University of Vic - Central University of Catalonia |
| BookMark | eNp9kEFPwjAYhhujiYD-AU9NvOih2nbrunobRMCEIAncm677huBYsR0x_HsnM3rz9H6H532_5Omj89rVgNANow-MUvkYGBdCEcoFYTQWgqgz1GNpkhClkvj89xbJJeqHsKU0EUrwHtpOl8P5_AlneLpZv5GlNZXJK8BDc4SwMTWew8Gbqo3m0_n3gDNroQJvGudbKECBXY3HG6gKvPBu7c1udxqYmAZw5r05Bnw3Xkyy-yt0UZoqwPVPDtBq_LwaTcnsdfIyymbEcskbUuRxATmLWJJyq5gEw3gprAAFJmW5LGJuDShWWikVFClVcVJCZCGNZcGKaIBuu9m9dx8HCI3euoOv24864jSmsRSKtRTvKOtdCB5KvfebnfFHzaj-Vqo7pbpVqk9KtWpLUVcKLVyvwf9N_9P6Agf6eic |
| Cites_doi | 10.1007/s11042-024-19953-8 10.1016/j.neucom.2024.128511 10.1016/j.csda.2019.106816 10.1080/01621459.2017.1285773 10.1109/MCSE.2010.69 10.1214/aoms/1177706645 10.1016/j.aei.2025.103132 10.1109/FPT.2017.8280160 10.1007/s10489-024-05898-w 10.1145/3529538.3529980 10.1109/JIOT.2024.3525020 10.3390/electronics13061074 10.1016/j.egyr.2025.01.039 10.1007/s12559-023-10242-4 10.1109/ICFPT52863.2021.9609847 10.1137/1036146 10.1109/TIE.2024.3370939 10.1007/s12559-023-10196-7 10.3389/fnins.2023.1291051 10.1145/3615661 10.1007/s12559-023-10187-8 10.3390/app15020688 10.1049/el.2015.3418 10.1145/1465482.1465560 10.1145/3296957.3173212 10.1007/s11265-021-01686-8 10.1007/s13389-024-00361-5 10.1016/j.future.2023.10.012 10.1093/biomet/52.3-4.591 10.1007/s10559-024-00655-w 10.1371/journal.pcbi.1011897 10.1145/42411.42415 10.1109/DAC18074.2021.9586137 10.1190/segam2020-3427239.1 10.1016/j.cub.2021.07.044 10.1109/TVT.2024.3397716 10.4064/cm-35-1-147-158 |
| ContentType | Journal Article |
| Copyright | The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2025 Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. Copyright Springer Nature B.V. Jun 2025 |
| Copyright_xml | – notice: The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2025 Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. – notice: Copyright Springer Nature B.V. Jun 2025 |
| DBID | AAYXX CITATION JQ2 |
| DOI | 10.1007/s12559-025-10455-9 |
| DatabaseName | CrossRef ProQuest Computer Science Collection |
| DatabaseTitle | CrossRef ProQuest Computer Science Collection |
| DatabaseTitleList | ProQuest Computer Science Collection |
| DeliveryMethod | fulltext_linktorsrc |
| Discipline | Psychology Computer Science |
| EISSN | 1866-9964 |
| ExternalDocumentID | 10_1007_s12559_025_10455_9 |
| GroupedDBID | 06C 06D 0R~ 0VY 1N0 203 29F 29~ 2JY 2KG 2VQ 2~H 30V 4.4 406 408 409 40D 40E 53G 67N 67Z 6NX 875 8TC 96X AAAVM AACDK AAFWJ AAHNG AAIAL AAJBT AAJKR AANXM AANZL AAPKM AARHV AARTL AASML AATNV AATVU AAUYE AAWCG AAYIU AAYQN AAYTO AAYZH AAZMS ABAKF ABBRH ABBXA ABDBE ABDZT ABECU ABFSG ABFTD ABFTV ABHLI ABHQN ABIVO ABJNI ABJOX ABKCH ABMQK ABPLI ABQBU ABSXP ABTEG ABTHY ABTKH ABTMW ABULA ABWNU ABXPI ACAOD ACDTI ACGFS ACHSB ACKNC ACMDZ ACMLO ACOKC ACOMO ACPIV ACSTC ACZOJ ADHHG ADHIR ADKFA ADKNI ADKPE ADRFC ADTPH ADURQ ADYFF ADZKW AEBTG AEFQL AEGNC AEJHL AEJRE AEKMD AEMSY AENEX AEOHA AEPYU AESKC AEVLU AEXYK AEZWR AFBBN AFDZB AFHIU AFKRA AFLOW AFOHR AFQWF AFWTZ AFZKB AGAYW AGDGC AGJBK AGMZJ AGQEE AGQMX AGRTI AGWZB AGYKE AHAVH AHBYD AHPBZ AHSBF AHWEU AHYZX AIAKS AIGIU AIIXL AILAN AITGF AIXLP AJBLW AJRNO AJZVZ AKMHD ALFXC ALMA_UNASSIGNED_HOLDINGS AMKLP AMXSW AMYLF AMYQR ANMIH AOCGG ARAPS ATHPR AUKKA AXYYD AYFIA BA0 BENPR BGLVJ BGNMA CAG CCPQU COF CSCUP DDRTE DNIVK DPUIP EBLON EBS EIOEI EJD EN4 ESBYG F5P FERAY FFXSO FIGPU FINBP FNLPD FRRFC FSGXE FYJPI G-Y G-Z GGCAI GGRSB GJIRD GQ7 GQ8 GXS H13 HCIFZ HF~ HLICF HMJXF HQYDN HRMNR HZ~ I0C IJ- IKXTQ IWAJR IXC IXD IZIGR I~X J-C J0Z JBSCW JCJTX JZLTJ K7- KOV KPH LLZTM M4Y NPVJJ NQJWS NU0 O9- O93 O9I O9J OAM PHGZM PHGZT PT4 QOR QOS R89 RLLFE ROL RSV S1Z S27 S3A S3B SBL SDH SHX SISQX SJYHP SNE SNPRN SNX SOHCF SOJ SPISZ SRMVM SSLCW SSXJD STPWE T13 TSG U2A U9L UG4 UOJIU UTJUX UZXMN VC2 VFIZW W48 WK8 Z45 ZMTXR ZOVNA ~A9 AAYXX ABRTQ AFFHD CITATION PQGLB JQ2 |
| ID | FETCH-LOGICAL-c272t-db4deb131682c917ea12f5c5e9ea81b7d42cae91fc779ed80946fe3ce847d1d3 |
| IEDL.DBID | RSV |
| ISICitedReferencesCount | 0 |
| ISICitedReferencesURI | http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=001494141700002&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D |
| ISSN | 1866-9956 |
| IngestDate | Wed Nov 26 13:51:08 EST 2025 Sat Nov 29 07:50:05 EST 2025 Thu Jun 26 01:51:21 EDT 2025 |
| IsPeerReviewed | true |
| IsScholarly | true |
| Issue | 3 |
| Keywords | Edge Computing Bio-inspired Neural Networks FPGA Bayesian Neural Networks Hardware Accelerator |
| Language | English |
| LinkModel | DirectLink |
| MergedId | FETCHMERGED-LOGICAL-c272t-db4deb131682c917ea12f5c5e9ea81b7d42cae91fc779ed80946fe3ce847d1d3 |
| Notes | ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 14 |
| ORCID | 0000-0002-6531-0769 |
| PQID | 3204047591 |
| PQPubID | 6623279 |
| ParticipantIDs | proquest_journals_3204047591 crossref_primary_10_1007_s12559_025_10455_9 springer_journals_10_1007_s12559_025_10455_9 |
| PublicationCentury | 2000 |
| PublicationDate | 20250600 2025-06-00 20250601 |
| PublicationDateYYYYMMDD | 2025-06-01 |
| PublicationDate_xml | – month: 6 year: 2025 text: 20250600 |
| PublicationDecade | 2020 |
| PublicationPlace | New York |
| PublicationPlace_xml | – name: New York |
| PublicationTitle | Cognitive computation |
| PublicationTitleAbbrev | Cogn Comput |
| PublicationYear | 2025 |
| Publisher | Springer US Springer Nature B.V |
| Publisher_xml | – name: Springer US – name: Springer Nature B.V |
| References | M Tasci (10455_CR13) 2025; 15 R Cai (10455_CR24) 2018; 53 10455_CR21 10455_CR22 Z Wang (10455_CR49) 2025; 65 X Wu (10455_CR25) 2024; 16 V Opanasenko (10455_CR2) 2024; 60 S López-Asunción (10455_CR18) 2024; 13 10455_CR30 10455_CR5 10455_CR31 Y Wan (10455_CR19) 2025; 55 10455_CR4 10455_CR3 D Yon (10455_CR1) 2021; 31 10455_CR39 10455_CR36 10455_CR37 10455_CR34 10455_CR33 JE Stone (10455_CR35) 2010; 12 R D’Agostino (10455_CR43) 1973; 60 10455_CR29 Z Wang (10455_CR10) 2024; 152 H Gou (10455_CR8) 2024; 16 JE Angus (10455_CR38) 1994; 36 10455_CR41 L Bai (10455_CR20) 2018; 65 J Grabinski (10455_CR27) 2022; 35 M Asiatici (10455_CR17) 2021; 93 10455_CR47 10455_CR48 10455_CR45 10455_CR46 10455_CR44 MH Alsharif (10455_CR11) 2025; 13 J Chen (10455_CR6) 2024; 16 K Ichikawa (10455_CR7) 2024; 20 10455_CR9 Z Liu (10455_CR15) 2024; 17 C Condo (10455_CR40) 2015; 51 SS Shapiro (10455_CR42) 1965; 52 10455_CR16 DM Blei (10455_CR32) 2017; 112 10455_CR14 A Rezaeezade (10455_CR26) 2024; 14 10455_CR12 H Li (10455_CR23) 2024; 17 A Le Coz (10455_CR28) 2024; 37 |
| References_xml | – ident: 10455_CR3 doi: 10.1007/s11042-024-19953-8 – ident: 10455_CR16 doi: 10.1016/j.neucom.2024.128511 – ident: 10455_CR34 doi: 10.1016/j.csda.2019.106816 – volume: 65 start-page: 1415 issue: 10 year: 2018 ident: 10455_CR20 publication-title: IEEE Transactions on Circuits and Systems II: Express Briefs. – volume: 37 start-page: 77686 year: 2024 ident: 10455_CR28 publication-title: Adv Neural Inf Process Syst. – volume: 112 start-page: 859 issue: 518 year: 2017 ident: 10455_CR32 publication-title: J Am Stat Assoc. doi: 10.1080/01621459.2017.1285773 – volume: 12 start-page: 66 issue: 3 year: 2010 ident: 10455_CR35 publication-title: Comput Sci & Eng. doi: 10.1109/MCSE.2010.69 – ident: 10455_CR39 doi: 10.1214/aoms/1177706645 – ident: 10455_CR41 – volume: 65 year: 2025 ident: 10455_CR49 publication-title: Adv Eng Inf. doi: 10.1016/j.aei.2025.103132 – ident: 10455_CR14 doi: 10.1109/FPT.2017.8280160 – volume: 55 start-page: 1 issue: 1 year: 2025 ident: 10455_CR19 publication-title: Appl Intell. doi: 10.1007/s10489-024-05898-w – ident: 10455_CR36 doi: 10.1145/3529538.3529980 – ident: 10455_CR45 – ident: 10455_CR9 doi: 10.1109/JIOT.2024.3525020 – volume: 13 start-page: 1074 issue: 6 year: 2024 ident: 10455_CR18 publication-title: Electronics. doi: 10.3390/electronics13061074 – ident: 10455_CR4 – volume: 13 start-page: 1739 year: 2025 ident: 10455_CR11 publication-title: Energy Rep. doi: 10.1016/j.egyr.2025.01.039 – volume: 16 start-page: 2471 issue: 5 year: 2024 ident: 10455_CR8 publication-title: Cogn Comput. doi: 10.1007/s12559-023-10242-4 – ident: 10455_CR22 doi: 10.1109/ICFPT52863.2021.9609847 – volume: 35 start-page: 39059 year: 2022 ident: 10455_CR27 publication-title: Adv Neural Inf Process Syst – ident: 10455_CR33 – volume: 36 start-page: 652 issue: 4 year: 1994 ident: 10455_CR38 publication-title: SIAM Rev. doi: 10.1137/1036146 – ident: 10455_CR12 – ident: 10455_CR48 doi: 10.1109/TIE.2024.3370939 – ident: 10455_CR44 – volume: 16 start-page: 121 issue: 1 year: 2024 ident: 10455_CR6 publication-title: Cogn Comput. doi: 10.1007/s12559-023-10196-7 – volume: 17 start-page: 1291051 year: 2024 ident: 10455_CR23 publication-title: Front Neurosci. doi: 10.3389/fnins.2023.1291051 – volume: 17 start-page: 1 issue: 1 year: 2024 ident: 10455_CR15 publication-title: ACM Trans Reconfigurable Technol Syst. doi: 10.1145/3615661 – volume: 16 start-page: 177 issue: 1 year: 2024 ident: 10455_CR25 publication-title: Cogn Comput. doi: 10.1007/s12559-023-10187-8 – volume: 15 start-page: 688 issue: 2 year: 2025 ident: 10455_CR13 publication-title: Appl Sci. doi: 10.3390/app15020688 – volume: 51 start-page: 2098 issue: 25 year: 2015 ident: 10455_CR40 publication-title: Electron Lett. doi: 10.1049/el.2015.3418 – ident: 10455_CR46 doi: 10.1145/1465482.1465560 – volume: 53 start-page: 476 year: 2018 ident: 10455_CR24 publication-title: SIGPLAN Not. doi: 10.1145/3296957.3173212 – volume: 93 start-page: 1405 year: 2021 ident: 10455_CR17 publication-title: J Signal Process Syst doi: 10.1007/s11265-021-01686-8 – volume: 14 start-page: 609 issue: 4 year: 2024 ident: 10455_CR26 publication-title: J Cryptographic Eng. doi: 10.1007/s13389-024-00361-5 – volume: 152 start-page: 55 year: 2024 ident: 10455_CR10 publication-title: Future Gen Comput Syst doi: 10.1016/j.future.2023.10.012 – volume: 52 start-page: 591 year: 1965 ident: 10455_CR42 publication-title: Biometrika doi: 10.1093/biomet/52.3-4.591 – volume: 60 start-page: 146 issue: 1 year: 2024 ident: 10455_CR2 publication-title: Cybern Syst Anal. doi: 10.1007/s10559-024-00655-w – volume: 20 issue: 3 year: 2024 ident: 10455_CR7 publication-title: PLoS Comput Biol. doi: 10.1371/journal.pcbi.1011897 – volume: 60 start-page: 613 year: 1973 ident: 10455_CR43 publication-title: Biometrika – ident: 10455_CR29 – ident: 10455_CR47 doi: 10.1145/42411.42415 – ident: 10455_CR21 doi: 10.1109/DAC18074.2021.9586137 – ident: 10455_CR30 doi: 10.1190/segam2020-3427239.1 – volume: 31 start-page: R1026 issue: 17 year: 2021 ident: 10455_CR1 publication-title: Curr Biol. doi: 10.1016/j.cub.2021.07.044 – ident: 10455_CR31 – ident: 10455_CR5 doi: 10.1109/TVT.2024.3397716 – ident: 10455_CR37 doi: 10.4064/cm-35-1-147-158 |
| SSID | ssj0065952 |
| Score | 2.343959 |
| Snippet | Traditional artificial neural networks have inherent overfitting problems and tend to produce overly confident predictions due to their reliance on point... |
| SourceID | proquest crossref springer |
| SourceType | Aggregation Database Index Database Publisher |
| StartPage | 100 |
| SubjectTerms | Artificial Intelligence Artificial neural networks Bayesian analysis Classification Computation Computation by Abstract Devices Computational Biology/Bioinformatics Computer Science Data processing Data transfer (computers) Edge computing Efficiency Field programmable gate arrays High acceleration Information sources Neural networks Normal distribution Optimization techniques Parallel processing Parameterization Pipelining (computers) Probability distribution Random numbers Statistical analysis |
| Title | HSBNN: A High-Scalable Bayesian Neural Networks Accelerator Based on Field Programmable Gate Arrays (FPGA) |
| URI | https://link.springer.com/article/10.1007/s12559-025-10455-9 https://www.proquest.com/docview/3204047591 |
| Volume | 17 |
| WOSCitedRecordID | wos001494141700002&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D |
| hasFullText | 1 |
| inHoldings | 1 |
| isFullTextHit | |
| isPrint | |
| journalDatabaseRights | – providerCode: PRVAVX databaseName: SpringerLINK Contemporary 1997-Present customDbUrl: eissn: 1866-9964 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0065952 issn: 1866-9956 databaseCode: RSV dateStart: 20090301 isFulltext: true titleUrlDefault: https://link.springer.com/search?facet-content-type=%22Journal%22 providerName: Springer Nature |
| link | http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwnV3NS8MwFH_o9ODF6VScTsnBg6KBNWubxlsnzh2kDDdkt5IlKSi4yTqF_fe-ZK1F0YOeEx7hveT33sv7AjhDGzgKhRJUtqWkvgzaNMKLQ43yMx5Kybh2U0vueZJE47EYFEVheZntXoYkHVJXxW7W-qV2_Cq6EEFAxTpsoLqL7MCGh-Fjib-2QZ6LcUZhSG3dZlEq8zONr-qosjG_hUWdtunV_3fOHdgurEsSr67DLqyZaQPq5eQGUjzkBmx94t5yD577w26SXJOY2KQPOkSp2Xoq0pVLY0ssiW3ggVSTVcZ4TmKlUFm5-Dxuyo0msynp2VQ4Mlile704AvZnDs8yl8ucnPcGd_HFPox6t6ObPi1GMFDFOFtQPfE1ormdbsUUenZGeiwLVGCEkWjwcu0zJY3wMsW5MDpCZzHMTEcZVHra050DqE1nU3MIRIYBriuuEUR8r60mHeYZNmGZ0G0d-aYJl6Ug0tdVo420aqlsWZoiS1PH0lQ0oVXKKi0eXZ52GCKS7V_oNeGqlE21_Du1o79tP4YtZsXr_mJaUFvM38wJbKr3xVM-P3WX8QPurtep |
| linkProvider | Springer Nature |
| linkToHtml | http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwnV1LSwMxEB60CvZitSpWq-bgQdHAbrqveKtirViXYov0tqRJFhRspVuF_nsn6a5F0YOeE4Ywk8wjM_MNwDH6wFHAJafCEYJ6wndohBeHaumlYSAEC5WdWtIJ4zgaDHg3bwrLimr3IiVpNfWi2c14v9SMX8UQwvcpX4YVDy2WQcx_6D0W-tcA5NkcZxQE1PRt5q0yP9P4ao4WPua3tKi1Nq3K_865Aeu5d0ma8-uwCUt6VIVKMbmB5A-5CuVPvTfbgud27zKOL0iTmKIP2kOpmX4qcilm2rRYEgPggVTjecV4RppSorGy-XnclGlFxiPSMqVwpDsv93qxBMzPHJ5lImYZOWl1b5qn29BvXfev2jQfwUAlC9mUqqGnUJub6VZMYmSnhctSX_qaa4EOb6g8JoXmbirDkGsVYbAYpLohNRo95arGDpRG45HeBSICH9dlqFCJeK4jhw3majZkKVeOijxdg7NCEMnrHGgjWUAqG5YmyNLEsjThNagXskryR5clDYYayeAXujU4L2SzWP6d2t7fth_BWrt_30k6t_HdPpSZEbX9l6lDaTp50wewKt-nT9nk0F7MDxOm2o0 |
| linkToPdf | http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwpV1LbxMxEB5BqKpcGghFDQ3gAwcqsJp19mVuCbAEEa0iJapyWzm2VwKJTZRNkfLvO-PdbSiih4qzrZE1Y8_DM98MwFv0geNQasnVQCnuq2DAY7w43Go_j0KlRGTc1JJplKbxcilnf6D4XbV7k5KsMA3UpanYXW5MfnkAvpEnzGkUK4YTQcDlY3jiUyE9xevzq0YXU7M8l--Mw5AThrOGzfybxl3TdPA3_0qROsuTdP7_zE_hpPY62ai6Js_gkS260GkmOrD6gXehfasP98_h52Q-TtOPbMSoGITPUZqEs2JjtbcEvWTU2AOpplUleclGWqMRc3l73FRaw9YFS6hEjs2qMrBfjgD92OFZtmpfsnfJ7Ovo4hQWyZfFpwmvRzNwLSKx42blG9TyNPVKaIz4rPJEHujASqvQEY6ML7Sy0st1FElrYgwiw9wOtUVjaDwzfAGtYl3YM2AqDHBdRwaVC0pQr4bCs2IlcmkGJvZtD943Qsk2VQOO7NBqmViaIUszx9JM9qDfyC2rH2OZDQVqKupr6PXgQyOnw_L91F4-bPsbOJ59TrLpt_T7ObQFSdp91_Shtdte21dwpH_vfpTb1-6O3gAASONx |
| openUrl | ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=HSBNN%3A+A+High-Scalable+Bayesian+Neural+Networks+Accelerator+Based+on+Field+Programmable+Gate+Arrays+%28FPGA%29&rft.jtitle=Cognitive+computation&rft.au=Liu%2C+Yinghao&rft.au=Zhang%2C+Hao&rft.au=Sun%2C+Zhe&rft.au=Duan%2C+Feng&rft.date=2025-06-01&rft.issn=1866-9956&rft.eissn=1866-9964&rft.volume=17&rft.issue=3&rft_id=info:doi/10.1007%2Fs12559-025-10455-9&rft.externalDBID=n%2Fa&rft.externalDocID=10_1007_s12559_025_10455_9 |
| thumbnail_l | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=1866-9956&client=summon |
| thumbnail_m | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=1866-9956&client=summon |
| thumbnail_s | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=1866-9956&client=summon |