1: \begin{abstract}

2:     Large-scale systems of linear equations arise in machine learning, medical imaging, sensor networks, and in many areas of data science.  When the scale of the systems are extreme, it is common for a fraction of the data or measurements to be corrupted. The Quantile Randomized Kaczmarz (QRK) method is known to converge on large-scale systems of linear equations $A\vx=\vb$ that are inconsistent due to \emph{static} corruptions in the measurement vector $\vb$. We prove that QRK converges even for systems corrupted by \emph{time-varying} perturbations. 

3:  Additionally, we prove that QRK converges up to a \emph{convergence horizon} on systems affected by time-varying noise and demonstrate that the noise affects only the convergence horizon of the method, and not the rate of convergence.  Finally, we utilize Markov's inequality to prove a lower bound on the probability that the largest entries of the QRK residual reveal the time-varying corruption in each iteration.  We present numerical experiments which illustrate our theoretical results. 

4: \end{abstract}

5: