Amazon Scholar, Amazon
2 papers at NeurIPS 2025
Methods based on Thompson Sampling for safe linear bandits that significantly improve computational costs while retaining regret and risk performance.