Coded Gradient Aggregation: A Tradeoff Between Communication Costs at Edge Nodes and at Helper Nodes

被引:8
作者
Sasidharan, Birenjith [1 ]
Thomas, Anoop [2 ]
机构
[1] Govt Engn Coll Barton Hill, Dept Elect & Commun Engn, Trivandrum 695035, Kerala, India
[2] Indian Inst Technol Bhubaneswar, Sch Elect Sci, Bhubaneswar 751013, Orissa, India
关键词
Costs; Codes; Servers; Encoding; Data privacy; Reliability; Distance learning; Gradient aggregation; learning at the edge; coded computing; locality;
D O I
10.1109/JSAC.2022.3142356
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
Increasing amount of data generated at edge nodes and quest for privacy have resulted in learning at the edge. Computations are performed at edge devices and outputs are communicated to a central node for updating the model. The edge nodes are available intermittently and are connected via low-bandwidth links. The edge nodes communicate local gradients to helper nodes, and these helpers forward messages to the central node after possible aggregation. Recently, schemes using repetition codes and maximum-distance-separable (MDS) codes, respectively known as aligned repetition coding (ARC) and aligned MDS coding (AMC) schemes, were proposed. It was observed that the communication cost at edge nodes becomes optimal in the AMC scheme, at the expense of an increased cost of communication incurred by helpers. An upper bound on the communication cost at helpers for the AMC scheme was known in literature. In this paper, a tradeoff between communication costs at edge nodes and at helper nodes is established with the help of newly proposed pyramid scheme. The scheme makes use of well-known class of pyramid codes, thus expanding the realm of application of locally repairable codes to distributed learning. The communication costs both at helper nodes and at edge nodes are exactly characterized. Using the developed technique, the exact communication cost at helper nodes can be computed for the AMC scheme as well. Next, we come up with a technique to improve the aggregation strategy of both pyramid and AMC schemes, that yields significant reduction in communication cost at helpers without changing parameters of the code used by edges. Finally, we present a greedy algorithm to improve the aggregation strategy of the ARC scheme, achieving significantly reduced communication cost at helpers.
引用
收藏
页码:761 / 772
页数:12
相关论文
共 30 条
[1]  
Acharya J, 2019, PR MACH LEARN RES, V97
[2]   Straggler Mitigation at Scale [J].
Aktas, Mehmet Fatih ;
Soljanin, Emina .
IEEE-ACM TRANSACTIONS ON NETWORKING, 2019, 27 (06) :2266-2279
[3]  
Alistarh D, 2017, ADV NEUR IN, V30
[4]   Practical Secure Aggregation for Privacy-Preserving Machine Learning [J].
Bonawitz, Keith ;
Ivanov, Vladimir ;
Kreuter, Ben ;
Marcedone, Antonio ;
McMahan, H. Brendan ;
Patel, Sarvar ;
Ramage, Daniel ;
Segal, Aaron ;
Seth, Karn .
CCS'17: PROCEEDINGS OF THE 2017 ACM SIGSAC CONFERENCE ON COMPUTER AND COMMUNICATIONS SECURITY, 2017, :1175-1191
[5]  
Cormen T. H., 2001, INTRO ALGORITHMS
[6]  
Dutta S, 2018, PR MACH LEARN RES, V84
[7]   On the Locality of Codeword Symbols [J].
Gopalan, Parikshit ;
Huang, Cheng ;
Simitci, Huseyin ;
Yekhanin, Sergey .
IEEE TRANSACTIONS ON INFORMATION THEORY, 2012, 58 (11) :6925-6934
[8]  
Halbawi W, 2018, IEEE INT SYMP INFO, P2027, DOI 10.1109/ISIT.2018.8437467
[9]   Pyramid codes: Flexible schemes to trade space for access efficiency in reliable data storage systems [J].
Huang, Cheng ;
Chen, Minghua ;
Li, Jin .
SIXTH IEEE INTERNATIONAL SYMPOSIUM ON NETWORK COMPUTING AND APPLICATIONS, PROCEEDINGS, 2007, :79-+
[10]  
Kadhe S, 2016, IEEE INT SYMP INFO, P435, DOI 10.1109/ISIT.2016.7541336