TSMOO: Solving Multi-Objective Experimentation with Constrained Thompson Sampling
This work introduces TSMOO (Thompson Sampling with Multi-Objective Optimization), a method that bridges the gap by combining multi-metric optimization with continuous learning in batch traffic allocation and outperforming both single-metric and elimination-based baselines.
K. C. Kalagarla, Wenyang Liu, Yi Liu et al.
· 0 citations