Unbiased Least Squares Regression via Averaged Stochastic Gradient Descent

View Full Paper

Ask AI

Bookmark

Authors

Discussion

Analysis reveals unbiased estimators reduce excess risk in least squares regression, suggesting improved accuracy with stochastic gradient descent.

Unbiased estimators improve accuracy in least squares regression, reducing expected excess risk.
Utilizing randomized multilevel Monte Carlo methods, we examine both biased and unbiased estimation techniques.
Our findings show significant confirmation through numerical experiments across regression tasks.
Implications for regression modeling include the benefits of average-start estimators in data analysis.

Nabil Kahale (2025) studied this question.

View Full Paper

Ask AI

Bookmark

Analysis reveals unbiased estimators reduce excess risk in least squares regression, suggesting improved accuracy with stochastic gradient descent.