Successive Approximation Coding for Distributed Matrix Multiplication

Coded distributed computing was recently introduced to mitigate the effect of stragglers on systems. This paper combines ideas approximate and coded further accelerate computation. We propose successive approximation coding (SAC) techniques that realize a tradeoff between accuracy speed, allowing system produce approximations increase in over time. If sufficient number compute nodes finish their tasks, SAC exactly recovers desired theoretically provide design guidelines for our techniques, numerically show achieves better accuracy-speed comparison with previous methods.