Coded Distributed Computing with Partial Recovery

TL;DR

This paper introduces a novel coded computation scheme that allows for partial recovery, balancing accuracy and speed in distributed computing, especially beneficial for iterative algorithms like machine learning.

Contribution

It proposes a new coded matrix-vector multiplication method called CCPR that reduces computation and decoding time by enabling approximate results, extending to general tasks with coded communication.

Findings

Reduces computation time and decoding complexity.

Enables trade-off between accuracy and latency.

Numerical simulations show improved convergence in linear regression.

Abstract

Coded computation techniques provide robustness against straggling workers in distributed computing. However, most of the existing schemes require exact provisioning of the straggling behaviour and ignore the computations carried out by straggling workers. Moreover, these schemes are typically designed to recover the desired computation results accurately, while in many machine learning and iterative optimization algorithms, faster approximate solutions are known to result in an improvement in the overall convergence time. In this paper, we first introduce a novel coded matrix-vector multiplication scheme, called coded computation with partial recovery (CCPR), which benefits from the advantages of both coded and uncoded computation schemes, and reduces both the computation time and the decoding complexity by allowing a trade-off between the accuracy and the speed of computation. We then extend this approach to distributed implementation of more general computation tasks by proposing a coded communication scheme with partial recovery, where the results of subtasks computed by the workers are coded before being communicated. Numerical simulations on a large linear regression task confirm the benefits of the proposed distributed computation scheme with partial recovery in terms of the trade-off between the computation accuracy and latency.