On the Rate of Gaussian Approximation for Linear Regression Problems
By: Marat Khusainov , Marina Sheshukova , Alain Durmus and more
Potential Business Impact:
Helps computers guess better with more data.
In this paper, we consider the problem of Gaussian approximation for the online linear regression task. We derive the corresponding rates for the setting of a constant learning rate and study the explicit dependence of the convergence rate upon the problem dimension $d$ and quantities related to the design matrix. When the number of iterations $n$ is known in advance, our results yield the rate of normal approximation of order $\sqrt{\log{n}/n}$, provided that the sample size $n$ is large enough.
Similar Papers
Gaussian Approximation for Two-Timescale Linear Stochastic Approximation
Machine Learning (Stat)
Makes computer learning more accurate and faster.
Online Linear Regression with Paid Stochastic Features
Machine Learning (CS)
Learns better by choosing how much to pay for cleaner data.
Uniform convergence for Gaussian kernel ridge regression
Machine Learning (Stat)
Makes computer learning more accurate and predictable.