PhD student, Seoul National University
1 paper at NeurIPS 2025
We propose a non-stationary structural causal bandit framework that models intervention propagation over time via temporal causal models, and introduce POMIS$^+$ to identify non-myopic strategies maximizing long-term rewards.