Distributed randomized Kaczmarz for the adversarial workers
This addresses the practical challenge of adversarial robustness in large-scale distributed computing, but it appears incremental as it builds on existing Kaczmarz methods.
The paper tackles the problem of making distributed least-squares methods robust to adversarial workers by proposing an iterative algorithm that uses simple statistics to guarantee convergence and learn adversarial distributions, with simulations showing high accuracy in tolerating adversary rates and identifying erroneous workers.
Developing large-scale distributed methods that are robust to the presence of adversarial or corrupted workers is an important part of making such methods practical for real-world problems. Here, we propose an iterative approach that is adversary-tolerant for least-squares problems. The algorithm utilizes simple statistics to guarantee convergence and is capable of learning the adversarial distributions. Additionally, the efficiency of the proposed method is shown in simulations in the presence of adversaries. The results demonstrate the great capability of such methods to tolerate different levels of adversary rates and to identify the erroneous workers with high accuracy.