Machine Learning at the Wireless Edge: Distributed Stochastic Gradient Descent Over-the-Air
We study federated machine learning (ML) at the wireless edge, where power- and bandwidth-limited wireless devices with local datasets carry out distributed stochastic gradient descent (DSGD) with the help of a parameter server (PS). Standard approaches assume separate computation and communication,...
Uložené v:
| Vydané v: | IEEE transactions on signal processing Ročník 68; s. 2155 - 2169 |
|---|---|
| Hlavní autori: | , |
| Médium: | Journal Article |
| Jazyk: | English |
| Vydavateľské údaje: |
New York
IEEE
2020
The Institute of Electrical and Electronics Engineers, Inc. (IEEE) |
| Predmet: | |
| ISSN: | 1053-587X, 1941-0476 |
| On-line prístup: | Získať plný text |
| Tagy: |
Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
|
| Abstract | We study federated machine learning (ML) at the wireless edge, where power- and bandwidth-limited wireless devices with local datasets carry out distributed stochastic gradient descent (DSGD) with the help of a parameter server (PS). Standard approaches assume separate computation and communication, where local gradient estimates are compressed and transmitted to the PS over orthogonal links. Following this digital approach, we introduce D-DSGD, in which the wireless devices employ gradient quantization and error accumulation, and transmit their gradient estimates to the PS over a multiple access channel (MAC). We then introduce a novel analog scheme, called A-DSGD, which exploits the additive nature of the wireless MAC for over-the-air gradient computation, and provide convergence analysis for this approach. In A-DSGD, the devices first sparsify their gradient estimates, and then project them to a lower dimensional space imposed by the available channel bandwidth. These projections are sent directly over the MAC without employing any digital code. Numerical results show that A-DSGD converges faster than D-DSGD thanks to its more efficient use of the limited bandwidth and the natural alignment of the gradient estimates over the channel. The improvement is particularly compelling at low power and low bandwidth regimes. We also illustrate for a classification problem that, A-DSGD is more robust to bias in data distribution across devices, while D-DSGD significantly outperforms other digital schemes in the literature. We also observe that both D-DSGD and A-DSGD perform better with the number of devices, showing their ability in harnessing the computation power of edge devices. |
|---|---|
| AbstractList | We study federated machine learning (ML) at the wireless edge, where power- and bandwidth-limited wireless devices with local datasets carry out distributed stochastic gradient descent (DSGD) with the help of a parameter server (PS). Standard approaches assume separate computation and communication, where local gradient estimates are compressed and transmitted to the PS over orthogonal links. Following this digital approach, we introduce D-DSGD, in which the wireless devices employ gradient quantization and error accumulation, and transmit their gradient estimates to the PS over a multiple access channel (MAC). We then introduce a novel analog scheme, called A-DSGD, which exploits the additive nature of the wireless MAC for over-the-air gradient computation, and provide convergence analysis for this approach. In A-DSGD, the devices first sparsify their gradient estimates, and then project them to a lower dimensional space imposed by the available channel bandwidth. These projections are sent directly over the MAC without employing any digital code. Numerical results show that A-DSGD converges faster than D-DSGD thanks to its more efficient use of the limited bandwidth and the natural alignment of the gradient estimates over the channel. The improvement is particularly compelling at low power and low bandwidth regimes. We also illustrate for a classification problem that, A-DSGD is more robust to bias in data distribution across devices, while D-DSGD significantly outperforms other digital schemes in the literature. We also observe that both D-DSGD and A-DSGD perform better with the number of devices, showing their ability in harnessing the computation power of edge devices. |
| Author | Gunduz, Deniz Mohammadi Amiri, Mohammad |
| Author_xml | – sequence: 1 givenname: Mohammad orcidid: 0000-0002-7343-6628 surname: Mohammadi Amiri fullname: Mohammadi Amiri, Mohammad email: mamiri@princeton.edu organization: Department of Electrical Engineering, Princeton University, Princeton, NJ, USA – sequence: 2 givenname: Deniz orcidid: 0000-0002-7725-395X surname: Gunduz fullname: Gunduz, Deniz email: d.gunduz@imperial.ac.uk organization: Department of Electrical and Electronic Engineering, Imperial College London, London, U.K |
| BookMark | eNp9kE1Lw0AQhhepYK3eBS8LnlP3Kx_rrbS1CpUKrSh4CJvJpN1Sk7q7Ffz3prR48ODpHZj3mYHnnHTqpkZCrjjrc8707WL-3BdMsL7QGddMnZAu14pHTKVJp51ZLKM4S9_OyLn3a8a4UjrpkvcnAytbI52icbWtl9QEGlZIX63DDXpPx-US7-jI-uBssQtY0nloYGV8sEAnzpQW60BH6GGfsy90UctHA-suyGllNh4vj9kjL_fjxfAhms4mj8PBNAKpdYhKyExWSc2SCiqRGog5SAZGg8QChSpSpRQwSE1cZYAaS5MpSBVLihKNQNkjN4e7W9d87tCHfN3sXN2-zIXUMU_iJFZtix1a4BrvHVb51tkP475zzvK9wrxVmO8V5keFLZL8QcAGE2xTB2fs5j_w-gBaRPz9026EjIX8AcW6gR4 |
| CODEN | ITPRED |
| CitedBy_id | crossref_primary_10_1109_JSAC_2022_3143217 crossref_primary_10_1109_JSAC_2024_3431572 crossref_primary_10_1109_TSP_2024_3351469 crossref_primary_10_1109_TSP_2025_3536023 crossref_primary_10_1109_TWC_2024_3457591 crossref_primary_10_1109_JSAC_2023_3242710 crossref_primary_10_1109_TWC_2023_3291877 crossref_primary_10_1007_s13369_021_06210_5 crossref_primary_10_1109_TITS_2022_3221388 crossref_primary_10_1002_ett_4458 crossref_primary_10_1016_j_micpro_2022_104596 crossref_primary_10_1109_ACCESS_2024_3474300 crossref_primary_10_1109_LWC_2022_3149783 crossref_primary_10_1109_TWC_2020_3038407 crossref_primary_10_1109_ACCESS_2023_3266297 crossref_primary_10_1109_COMST_2023_3264649 crossref_primary_10_1109_TWC_2020_3043787 crossref_primary_10_1109_JIOT_2023_3349176 crossref_primary_10_1109_LCOMM_2023_3272844 crossref_primary_10_1109_TCOMM_2022_3175175 crossref_primary_10_1109_JSAC_2023_3242722 crossref_primary_10_1109_JIOT_2022_3164914 crossref_primary_10_1109_TSP_2022_3186526 crossref_primary_10_1007_s11280_021_00903_5 crossref_primary_10_1109_JSAC_2022_3229428 crossref_primary_10_1109_ACCESS_2020_3001752 crossref_primary_10_1109_JSAC_2023_3242727 crossref_primary_10_1109_MCOM_101_2100138 crossref_primary_10_1109_TWC_2022_3185156 crossref_primary_10_1109_TIT_2025_3542673 crossref_primary_10_1109_TVT_2021_3135541 crossref_primary_10_1109_TWC_2023_3268742 crossref_primary_10_1109_TMC_2021_3123195 crossref_primary_10_1109_ACCESS_2023_3291023 crossref_primary_10_1109_JIOT_2024_3376821 crossref_primary_10_1080_03610926_2025_2474629 crossref_primary_10_1109_TNNLS_2022_3170944 crossref_primary_10_1109_TVT_2022_3161099 crossref_primary_10_1109_TGCN_2022_3186538 crossref_primary_10_1109_TWC_2023_3313968 crossref_primary_10_1109_TWC_2020_3000241 crossref_primary_10_1109_COMST_2023_3302474 crossref_primary_10_1109_JIOT_2023_3340014 crossref_primary_10_1109_TCOMM_2021_3081746 crossref_primary_10_1109_TWC_2024_3432758 crossref_primary_10_1166_jmihi_2021_3684 crossref_primary_10_1109_TWC_2023_3288122 crossref_primary_10_1109_TWC_2024_3501297 crossref_primary_10_1109_TCC_2025_3530681 crossref_primary_10_1109_TSP_2021_3090323 crossref_primary_10_1109_JSAC_2022_3223661 crossref_primary_10_1109_TWC_2021_3086116 crossref_primary_10_1109_TNNLS_2024_3394861 crossref_primary_10_1109_JIOT_2022_3164339 crossref_primary_10_1109_TCCN_2021_3074908 crossref_primary_10_1109_TWC_2023_3316788 crossref_primary_10_1016_j_ins_2022_11_157 crossref_primary_10_1007_s11432_020_3251_9 crossref_primary_10_1109_JIOT_2021_3111624 crossref_primary_10_1109_MSP_2021_3125282 crossref_primary_10_1007_s10489_022_03401_x crossref_primary_10_1109_TKDE_2023_3332770 crossref_primary_10_1109_JSAC_2022_3180799 crossref_primary_10_1109_LWC_2021_3075029 crossref_primary_10_1016_j_jfranklin_2025_107510 crossref_primary_10_1002_ett_4661 crossref_primary_10_1007_s10773_021_04867_0 crossref_primary_10_3390_s21134586 crossref_primary_10_1109_TSP_2023_3290327 crossref_primary_10_1109_JIOT_2021_3132031 crossref_primary_10_1109_ACCESS_2022_3195056 crossref_primary_10_1007_s00521_025_11248_9 crossref_primary_10_1109_LWC_2021_3099136 crossref_primary_10_1109_TWC_2024_3366547 crossref_primary_10_1080_09540091_2021_1936455 crossref_primary_10_1109_COMST_2020_3007787 crossref_primary_10_1109_TCCN_2024_3419039 crossref_primary_10_1109_TWC_2021_3099505 crossref_primary_10_3390_s23010040 crossref_primary_10_1002_int_23056 crossref_primary_10_1109_TNET_2023_3318474 crossref_primary_10_3390_jmse10060743 crossref_primary_10_1109_TWC_2021_3103874 crossref_primary_10_1109_LCOMM_2020_3019312 crossref_primary_10_1109_TWC_2022_3173262 crossref_primary_10_1109_JSAC_2022_3223408 crossref_primary_10_1109_TWC_2022_3188502 crossref_primary_10_1109_TSP_2025_3592678 crossref_primary_10_1109_JIOT_2022_3188544 crossref_primary_10_1109_JIOT_2023_3285868 crossref_primary_10_1109_TCOMM_2021_3124961 crossref_primary_10_1109_TVT_2022_3210335 crossref_primary_10_1007_s10994_023_06332_x crossref_primary_10_1109_TSP_2023_3284381 crossref_primary_10_1109_TGCN_2021_3125232 crossref_primary_10_1109_LSP_2023_3321561 crossref_primary_10_1109_TWC_2022_3231088 crossref_primary_10_1109_LCOMM_2025_3567387 crossref_primary_10_1109_TWC_2021_3104834 crossref_primary_10_1109_COMST_2021_3086014 crossref_primary_10_1109_TGCN_2022_3173420 crossref_primary_10_1109_TSP_2023_3284376 crossref_primary_10_1109_JIOT_2021_3131407 crossref_primary_10_1109_JIOT_2022_3168066 crossref_primary_10_1109_TWC_2023_3263148 crossref_primary_10_1109_TCOMM_2022_3162576 crossref_primary_10_1109_TMC_2024_3427709 crossref_primary_10_1109_TSP_2022_3185897 crossref_primary_10_1109_TGCN_2021_3126795 crossref_primary_10_1109_TSP_2025_3585002 crossref_primary_10_1080_0954898X_2025_2513690 crossref_primary_10_1109_TWC_2024_3368197 crossref_primary_10_1109_TWC_2024_3407822 crossref_primary_10_1109_LCOMM_2022_3203581 crossref_primary_10_1109_TWC_2023_3347697 crossref_primary_10_1109_JIOT_2022_3218008 crossref_primary_10_1109_TPDS_2022_3149787 crossref_primary_10_1109_TWC_2022_3209190 crossref_primary_10_1109_TMC_2023_3292898 crossref_primary_10_1109_TWC_2022_3181214 crossref_primary_10_1109_TWC_2021_3052681 crossref_primary_10_1109_JSEN_2020_3023661 crossref_primary_10_1109_TCCN_2022_3181032 crossref_primary_10_3390_s23187840 crossref_primary_10_1109_TWC_2020_3039309 crossref_primary_10_1109_JSAC_2021_3118353 crossref_primary_10_3390_s21155124 crossref_primary_10_1109_TWC_2023_3329375 crossref_primary_10_1109_TWC_2021_3138394 crossref_primary_10_1109_JSAC_2024_3413988 crossref_primary_10_1109_TSP_2024_3454986 crossref_primary_10_1109_TWC_2024_3453392 crossref_primary_10_1109_ACCESS_2022_3231039 crossref_primary_10_1109_JIOT_2021_3096570 crossref_primary_10_1109_TCOMM_2023_3283790 crossref_primary_10_1109_ACCESS_2022_3141913 crossref_primary_10_1109_TNSE_2022_3196463 crossref_primary_10_1109_TWC_2021_3065920 crossref_primary_10_1109_TWC_2023_3235894 crossref_primary_10_1109_TWC_2021_3092075 crossref_primary_10_1109_TWC_2023_3262350 crossref_primary_10_1109_ACCESS_2025_3564484 crossref_primary_10_1109_JSAC_2021_3118348 crossref_primary_10_1109_LSP_2024_3400916 crossref_primary_10_1109_JIOT_2023_3285937 crossref_primary_10_1109_TII_2022_3170900 crossref_primary_10_1109_JSAC_2021_3118346 crossref_primary_10_1109_JSTSP_2024_3463505 crossref_primary_10_1109_TWC_2021_3088910 crossref_primary_10_1109_TWC_2022_3201207 crossref_primary_10_3390_electronics12143021 crossref_primary_10_1190_geo2023_0751_1 crossref_primary_10_3390_electronics11152464 crossref_primary_10_1016_j_future_2021_07_010 crossref_primary_10_1109_TSP_2022_3214122 crossref_primary_10_1109_TWC_2022_3208198 crossref_primary_10_1016_j_future_2022_01_013 crossref_primary_10_1109_TWC_2023_3326438 crossref_primary_10_1109_LSP_2022_3154653 crossref_primary_10_1109_TSP_2023_3329974 crossref_primary_10_1109_JIOT_2024_3523947 crossref_primary_10_1109_TGCN_2023_3239373 crossref_primary_10_1109_JIOT_2022_3230452 crossref_primary_10_1109_TNET_2024_3444602 crossref_primary_10_1109_TWC_2023_3317129 crossref_primary_10_1109_JSAC_2021_3126076 crossref_primary_10_1109_TCCN_2022_3140788 crossref_primary_10_1109_TWC_2022_3192550 crossref_primary_10_1109_JSAC_2021_3126078 crossref_primary_10_1109_COMST_2021_3058573 crossref_primary_10_1109_JSTSP_2022_3156756 crossref_primary_10_1016_j_image_2024_117199 crossref_primary_10_1109_LWC_2023_3256006 crossref_primary_10_3390_su151813951 crossref_primary_10_1109_OJCOMS_2024_3521651 crossref_primary_10_1109_TWC_2021_3098716 crossref_primary_10_1109_ACCESS_2025_3543441 crossref_primary_10_1109_JSAC_2020_3041379 crossref_primary_10_1109_JIOT_2022_3165268 crossref_primary_10_1109_TGCN_2023_3310569 crossref_primary_10_1109_TWC_2025_3540455 crossref_primary_10_1109_TCOMM_2023_3277882 crossref_primary_10_1109_MCOM_001_2000394 crossref_primary_10_1109_JIOT_2024_3360190 crossref_primary_10_1631_FITEE_2100538 crossref_primary_10_1109_TCOMM_2021_3114791 crossref_primary_10_1109_TMC_2024_3411295 crossref_primary_10_1109_JIOT_2024_3376771 crossref_primary_10_1016_j_eswa_2025_126474 crossref_primary_10_1109_JIOT_2022_3170449 crossref_primary_10_1109_TSP_2025_3534685 crossref_primary_10_1109_TNSE_2025_3528982 crossref_primary_10_1109_TWC_2021_3125798 crossref_primary_10_1109_TWC_2021_3087594 crossref_primary_10_1109_TVT_2025_3533576 crossref_primary_10_1109_TSP_2024_3358615 crossref_primary_10_1016_j_phycom_2023_102201 crossref_primary_10_1109_TGCN_2022_3186439 crossref_primary_10_1109_TWC_2021_3130111 crossref_primary_10_1016_j_sysarc_2022_102673 crossref_primary_10_1109_JIOT_2024_3391923 crossref_primary_10_1109_TCOMM_2021_3051397 crossref_primary_10_1109_TWC_2023_3315962 crossref_primary_10_1109_JSAC_2021_3118428 crossref_primary_10_1109_JSAC_2021_3126060 crossref_primary_10_1109_TMC_2021_3137017 crossref_primary_10_1109_MCOM_001_2200962 crossref_primary_10_1109_TCOMM_2022_3198433 crossref_primary_10_1109_TWC_2022_3185074 crossref_primary_10_1109_TMC_2025_3547148 crossref_primary_10_3390_a17020052 crossref_primary_10_1016_j_cag_2023_10_016 crossref_primary_10_1109_LWC_2021_3095169 crossref_primary_10_1109_JIOT_2024_3438082 crossref_primary_10_1109_TCOMM_2022_3153488 crossref_primary_10_1109_TBDATA_2024_3403381 crossref_primary_10_1109_TWC_2024_3425732 crossref_primary_10_1109_TSP_2021_3106115 crossref_primary_10_1109_TWC_2021_3128392 crossref_primary_10_1109_JIOT_2024_3352280 crossref_primary_10_1109_JSAC_2021_3118436 crossref_primary_10_1109_TII_2024_3431100 crossref_primary_10_1109_COMST_2022_3218527 crossref_primary_10_1109_TWC_2020_3008091 crossref_primary_10_1109_TWC_2024_3487986 crossref_primary_10_1109_JIOT_2020_3002925 crossref_primary_10_1007_s00371_023_03043_1 crossref_primary_10_1109_LWC_2020_2996194 crossref_primary_10_1109_TII_2022_3145010 crossref_primary_10_1109_TSP_2021_3079808 crossref_primary_10_1109_JSAC_2021_3118408 crossref_primary_10_1109_JSAC_2021_3118406 crossref_primary_10_1109_JSAC_2021_3118402 crossref_primary_10_1109_JSTSP_2021_3139231 crossref_primary_10_3390_rs16163070 crossref_primary_10_1109_TMC_2022_3216837 crossref_primary_10_1109_LSP_2024_3384077 crossref_primary_10_1109_TSP_2024_3460690 crossref_primary_10_1109_JSAC_2020_3036948 crossref_primary_10_1109_TSP_2025_3540655 crossref_primary_10_1109_TNNLS_2024_3362974 crossref_primary_10_1109_TCOMM_2025_3535878 crossref_primary_10_1109_MWC_011_2000501 crossref_primary_10_1109_JIOT_2023_3296493 crossref_primary_10_1109_TWC_2022_3225812 crossref_primary_10_1109_TMC_2023_3304652 crossref_primary_10_1109_JSAC_2021_3118412 crossref_primary_10_1109_MBITS_2023_3322667 crossref_primary_10_1109_JSAC_2021_3118410 crossref_primary_10_1109_MWC_005_2300016 |
| Cites_doi | 10.1073/pnas.0909892106 10.1109/TWC.2020.2974748 10.18653/v1/D17-1045 10.1109/TIT.2017.2649460 10.1109/IJCNN.2019.8852172 10.1109/PIMRC.2019.8904164 10.1109/ISIT.2019.8849334 10.1109/TWC.2019.2961673 10.3389/fams.2018.00062 10.1109/LCOMM.2018.2877316 10.1109/TIT.2010.2094817 10.1214/aop/1176989118 10.1109/GlobalSIP45357.2019.8969185 10.1109/ASPDAC.2018.8297378 10.1109/TWC.2019.2946245 10.1109/TCOMM.2013.072913.120815 10.1145/3295500.3356222 10.1109/TIT.2018.2882177 10.1109/TIT.2009.2025566 10.1109/SPAWC.2019.8815402 |
| ContentType | Journal Article |
| Copyright | Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2020 |
| Copyright_xml | – notice: Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2020 |
| DBID | 97E RIA RIE AAYXX CITATION 7SC 7SP 8FD JQ2 L7M L~C L~D |
| DOI | 10.1109/TSP.2020.2981904 |
| DatabaseName | IEEE Xplore (IEEE) IEEE All-Society Periodicals Package (ASPP) Online IEEE Electronic Library (IEL) CrossRef Computer and Information Systems Abstracts Electronics & Communications Abstracts Technology Research Database ProQuest Computer Science Collection Advanced Technologies Database with Aerospace Computer and Information Systems Abstracts Academic Computer and Information Systems Abstracts Professional |
| DatabaseTitle | CrossRef Technology Research Database Computer and Information Systems Abstracts – Academic Electronics & Communications Abstracts ProQuest Computer Science Collection Computer and Information Systems Abstracts Advanced Technologies Database with Aerospace Computer and Information Systems Abstracts Professional |
| DatabaseTitleList | Technology Research Database |
| Database_xml | – sequence: 1 dbid: RIE name: IEEE Electronic Library (IEL) url: https://ieeexplore.ieee.org/ sourceTypes: Publisher |
| DeliveryMethod | fulltext_linktorsrc |
| Discipline | Engineering |
| EISSN | 1941-0476 |
| EndPage | 2169 |
| ExternalDocumentID | 10_1109_TSP_2020_2981904 9042352 |
| Genre | orig-research |
| GrantInformation_xml | – fundername: BEACON grantid: 677854 – fundername: European Research Council funderid: 10.13039/501100000781 |
| GroupedDBID | -~X .DC 0R~ 29I 3EH 4.4 53G 5GY 5VS 6IK 85S 97E AAJGR AARMG AASAJ AAWTH ABAZT ABFSI ABQJQ ABVLG ACGFO ACIWK ACKIV ACNCT AENEX AETIX AGQYO AGSQL AHBIQ AI. AIBXA AJQPL AKJIK AKQYR ALLEH ALMA_UNASSIGNED_HOLDINGS ASUFR ATWAV BEFXN BFFAM BGNUA BKEBE BPEOZ CS3 E.L EBS EJD F5P HZ~ H~9 ICLAB IFIPE IFJZH IPLJI JAVBF LAI MS~ O9- OCL P2P RIA RIE RNS TAE TN5 VH1 AAYXX CITATION 7SC 7SP 8FD JQ2 L7M L~C L~D |
| ID | FETCH-LOGICAL-c399t-dc8a8f3906fcf27ac51c30ca9c3ebe24b7444c0c7a5f8ce9eda84c7406bdea2e3 |
| IEDL.DBID | RIE |
| ISICitedReferencesCount | 71 |
| ISICitedReferencesURI | http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=000489100301105&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D |
| ISSN | 1053-587X |
| IngestDate | Mon Jun 30 10:19:05 EDT 2025 Sat Nov 29 04:10:51 EST 2025 Tue Nov 18 20:53:12 EST 2025 Wed Aug 27 02:42:21 EDT 2025 |
| IsDoiOpenAccess | false |
| IsOpenAccess | true |
| IsPeerReviewed | true |
| IsScholarly | true |
| Language | English |
| License | https://ieeexplore.ieee.org/Xplorehelp/downloads/license-information/IEEE.html https://doi.org/10.15223/policy-029 https://doi.org/10.15223/policy-037 |
| LinkModel | DirectLink |
| MergedId | FETCHMERGED-LOGICAL-c399t-dc8a8f3906fcf27ac51c30ca9c3ebe24b7444c0c7a5f8ce9eda84c7406bdea2e3 |
| Notes | ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 14 |
| ORCID | 0000-0002-7343-6628 0000-0002-7725-395X |
| OpenAccessLink | http://hdl.handle.net/11380/1217809 |
| PQID | 2395165654 |
| PQPubID | 85478 |
| PageCount | 15 |
| ParticipantIDs | crossref_citationtrail_10_1109_TSP_2020_2981904 proquest_journals_2395165654 crossref_primary_10_1109_TSP_2020_2981904 ieee_primary_9042352 |
| PublicationCentury | 2000 |
| PublicationDate | 2020-00-00 |
| PublicationDateYYYYMMDD | 2020-01-01 |
| PublicationDate_xml | – year: 2020 text: 2020-00-00 |
| PublicationDecade | 2020 |
| PublicationPlace | New York |
| PublicationPlace_xml | – name: New York |
| PublicationTitle | IEEE transactions on signal processing |
| PublicationTitleAbbrev | TSP |
| PublicationYear | 2020 |
| Publisher | IEEE The Institute of Electrical and Electronics Engineers, Inc. (IEEE) |
| Publisher_xml | – name: IEEE – name: The Institute of Electrical and Electronics Engineers, Inc. (IEEE) |
| References | ref35 ref34 samarakoon (ref12) 2018 ref37 ref36 ref31 wang (ref15) 2018 alistarh (ref23) 2018 ref30 ref33 ref32 nishio (ref11) 2018 zhou (ref14) 2018 alistarh (ref2) 2017 ref18 strom (ref19) 2015 johnson (ref39) 0 park (ref1) 2018 strom (ref38) 0 konecny (ref5) 2016 lecun (ref45) 1998 konecny (ref10) 2015 mcmahan (ref7) 2017 ref47 wen (ref13) 2017 ref20 ref42 ref41 smith (ref8) 2017 ref22 ref21 seide (ref4) 0 tsuzuku (ref24) 2018 stich (ref25) 2018 ref28 ref27 ref29 sa (ref44) 0 ref9 lin (ref26) 2018 mood (ref43) 1974 wu (ref17) 2018; 80 lin (ref3) 2018 ref40 kingma (ref46) 2017 bernstein (ref16) 2018; 80 mcmahan (ref6) 0 |
| References_xml | – year: 2018 ident: ref11 article-title: Client selection for federated learning with heterogeneous resources in mobile edge – year: 2016 ident: ref5 article-title: Federated learning: Strategies for improving communication efficiency publication-title: Proc NIPS Workshop Private Multi-Party Mach Learn – year: 2017 ident: ref7 article-title: Federated learning: Collaborative machine learning without centralized training data – ident: ref31 doi: 10.1073/pnas.0909892106 – ident: ref34 doi: 10.1109/TWC.2020.2974748 – start-page: 315 year: 0 ident: ref39 article-title: Accelerating stochastic gradient descent using predictive variance reduction publication-title: Proc Int Conf Neural Inf Process – year: 2018 ident: ref3 article-title: Deep gradient compression: Reducing the communication bandwidth for distributed training publication-title: Proc Int Conf Learn Representations – ident: ref20 doi: 10.18653/v1/D17-1045 – start-page: 2674 year: 0 ident: ref44 article-title: Taming the wild: A unified analysis of hogwild-style algorithms publication-title: Proc Int Conf Neural Inf Process – year: 0 ident: ref6 article-title: Communication-efficient learning of deep networks from decentralized data publication-title: Proc 20th Int Conf Artif Intell Statist – year: 2017 ident: ref46 article-title: Adam: A method for stochastic optimization – year: 2017 ident: ref13 article-title: TernGrad: Ternary gradients to reduce communication in distributed deep learning publication-title: Proc Conf Neural Inf Process Syst – ident: ref41 doi: 10.1109/TIT.2017.2649460 – ident: ref21 doi: 10.1109/IJCNN.2019.8852172 – year: 2015 ident: ref10 article-title: Federated optimization: Distributed optimization beyond the datacenter – ident: ref36 doi: 10.1109/PIMRC.2019.8904164 – ident: ref29 doi: 10.1109/ISIT.2019.8849334 – ident: ref33 doi: 10.1109/TWC.2019.2961673 – year: 2018 ident: ref23 article-title: The convergence of sparsified gradient methods publication-title: Proc Conf Neural Inf Process Syst – ident: ref9 doi: 10.3389/fams.2018.00062 – ident: ref30 doi: 10.1109/LCOMM.2018.2877316 – ident: ref40 doi: 10.1109/TIT.2010.2094817 – year: 1998 ident: ref45 article-title: The MNIST database of handwritten digits – ident: ref47 doi: 10.1214/aop/1176989118 – ident: ref37 doi: 10.1109/GlobalSIP45357.2019.8969185 – ident: ref18 doi: 10.1109/ASPDAC.2018.8297378 – year: 1974 ident: ref43 publication-title: Introduction to the Theory of Statistics – year: 2018 ident: ref26 article-title: Don't use large mini-batches, use local SGD – volume: 80 start-page: 5325 year: 2018 ident: ref17 article-title: Error compensated quantized SGD and its applications to large-scale distributed optimization publication-title: Proc Int Conf Mach Learn – year: 2015 ident: ref19 article-title: Scalable distributed DNN training using commodity gpu cloud computing publication-title: Proc Conf Int Speech Commun Assoc (INTERSPEECH) – year: 2017 ident: ref2 article-title: QSGD: Communication-efficient SGD via gradient quantization and encoding publication-title: Proc Conf Neural Inf Process Syst – year: 2018 ident: ref12 article-title: Distributed federated learning for ultra-reliable low-latency vehicular communications – start-page: 1488 year: 0 ident: ref38 article-title: Scalable distributed DNN training using commodity GPU cloud computing publication-title: Proc INTERSPEECH – ident: ref32 doi: 10.1109/TWC.2019.2946245 – year: 2018 ident: ref15 article-title: ATOMO: Communication-efficient learning via atomic sparsification publication-title: Proc Conf Neural Inf Process Syst – year: 2018 ident: ref24 article-title: Variance-based gradient compression for efficient distributed deep learning – ident: ref28 doi: 10.1109/TCOMM.2013.072913.120815 – ident: ref22 doi: 10.1145/3295500.3356222 – volume: 80 start-page: 560 year: 2018 ident: ref16 article-title: signSGD: Compressed optimisation for non-convex problems publication-title: Proc Int Conf Mach Learn – ident: ref42 doi: 10.1109/TIT.2018.2882177 – year: 2018 ident: ref25 article-title: Local SGD converges fast and communicates little – year: 2018 ident: ref1 article-title: Wireless network intelligence at the edge – year: 2018 ident: ref14 article-title: DoReFa-Net: Training low bitwidth convolutional neural networks with low bitwidth gradients – start-page: 1058 year: 0 ident: ref4 article-title: 1-bit stochastic gradient descent and its application to data-parallel distributed training of speech DNNs publication-title: Proc INTERSPEECH – ident: ref27 doi: 10.1109/TIT.2009.2025566 – year: 2017 ident: ref8 article-title: Federated multi-task learning publication-title: Proc Conf Neural Inf Process Syst – ident: ref35 doi: 10.1109/SPAWC.2019.8815402 |
| SSID | ssj0014496 |
| Score | 2.7162306 |
| Snippet | We study federated machine learning (ML) at the wireless edge, where power- and bandwidth-limited wireless devices with local datasets carry out distributed... |
| SourceID | proquest crossref ieee |
| SourceType | Aggregation Database Enrichment Source Index Database Publisher |
| StartPage | 2155 |
| SubjectTerms | Approximate message passing (AMP) Bandwidth Bandwidths Channel estimation Computation Convergence Electronic devices Estimates federated learning (FL) Machine learning Machine learning algorithms over-the-air computation Performance evaluation Power management Quantization (signal) Robustness (mathematics) stochastic gradient descent (SGD) Wireless communication Wireless sensor networks |
| Title | Machine Learning at the Wireless Edge: Distributed Stochastic Gradient Descent Over-the-Air |
| URI | https://ieeexplore.ieee.org/document/9042352 https://www.proquest.com/docview/2395165654 |
| Volume | 68 |
| WOSCitedRecordID | wos000489100301105&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: PRVIEE databaseName: IEEE Electronic Library (IEL) customDbUrl: eissn: 1941-0476 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0014496 issn: 1053-587X databaseCode: RIE dateStart: 19910101 isFulltext: true titleUrlDefault: https://ieeexplore.ieee.org/ providerName: IEEE |
| link | http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV3PT8IwFG6QeNCDv9CIounBi4mF0XVr642I6EGRBDQkHpby1iEJATOGf79tN4hGY-Jth3ZZ3te999r3-n0IXVABngqFJE2pBGFJkBAB3Cc6UYGQILhwKhEvD7zbFcOh7JXQ1foujNbaNZ_pun10tfx4Dkt7VNaQtokjMA53g_Mwv6u1rhgw5rS4TLrgk0Dw4aok6cnGoN8zG0Hq1am08Y99C0FOU-WHI3bRpbP7v-_aQztFFolbOez7qKRnB2j7C7dgBb0-ujZJjQsG1TFWGTbZHrb9rlPj3_BtPNbXuG2Zc63olY5xP5vDm7LMzfgudb1gGW7nfE_4yax5YuaT1iQ9RM-d28HNPSmUFAiYBCQjMQglEl96YQIJ5QqCJvgeKAm-AZGyEWeMgQdcBYkALXWsBANugv0o1opq_wiVZ_OZPkY4TEJhOfy0DCWDpq88wajlPeMGcxhBFTVWxo2goBm3ahfTyG03PBkZOCILR1TAUUWX6xnvOcXGH2Mr1vzrcYXlq6i2wi8q_sFFRH2TPZp0NWAnv886RVv23fmBSg2Vs3Spz9AmfGSTRXrultcnoZvM1Q |
| linkProvider | IEEE |
| linkToHtml | http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV1LS8QwEB5EBfXgW1yfOXgRjJtN0ybxJr5xXQVXWfBQstNUBdmVtfr7TdLuoiiCtx4SWuZLZyaZyfcB7HCFzCRK04Y2ioo8zqlCGVGbm1hpVFIFlYj7pmy1VKejb8Zgb3QXxlobms_svn8Mtfysj-_-qKyufRNH7BzuRCwEZ-VtrVHNQIigxuUShojGSnaGRUmm6-3bG7cV5Gyfax8BxbcgFFRVfrjiEF9O5_73ZfMwW-WR5LAEfgHGbG8RZr6wCy7Bw1VolLSk4lB9JKYgLt8jvuP1xXk4cpI92gNy7LlzveyVzcht0ccn47mbydkgdIMV5LhkfCLXbtVTN58ePg-W4e70pH10TistBYouBSlohsqoPNIsyTHn0mDcwIih0Rg5GLnoSiEEMpQmzhVabTOjBEoX7ruZNdxGKzDe6_fsKpAkT5Rn8bM60QIbkWFKcM98Jh3q2MUa1IfGTbEiGvd6Fy9p2HAwnTo4Ug9HWsFRg93RjNeSZOOPsUve_KNxleVrsDHEL63-wreURy5_dAlrLNZ-n7UNU-ftq2bavGhdrsO0f095vLIB48Xg3W7CJH4Uz2-DrbDUPgFyzdAc |
| 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=Machine+Learning+at+the+Wireless+Edge%3A+Distributed+Stochastic+Gradient+Descent+Over-the-Air&rft.jtitle=IEEE+transactions+on+signal+processing&rft.au=Mohammadi+Amiri%2C+Mohammad&rft.au=Gunduz%2C+Deniz&rft.date=2020&rft.issn=1053-587X&rft.eissn=1941-0476&rft.volume=68&rft.spage=2155&rft.epage=2169&rft_id=info:doi/10.1109%2FTSP.2020.2981904&rft.externalDBID=n%2Fa&rft.externalDocID=10_1109_TSP_2020_2981904 |
| thumbnail_l | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=1053-587X&client=summon |
| thumbnail_m | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=1053-587X&client=summon |
| thumbnail_s | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=1053-587X&client=summon |