Thresholded Lasso achieves the minimax cumulative estimation error s0(log d + log T) in sequential sparse regression, and a three-stage algorithm using it is nearly minimax optimal for high-dimensional linear bandits.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
math.ST 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Minimax Rate-Optimal Algorithms for High-Dimensional Stochastic Linear Bandits
Thresholded Lasso achieves the minimax cumulative estimation error s0(log d + log T) in sequential sparse regression, and a three-stage algorithm using it is nearly minimax optimal for high-dimensional linear bandits.