6 problems

Time-varying perturbation conjecture. Optimal regret could still be achieved using

The procedure discussed above concerns inference after adaptive data collection in a contextual bandit with stopping time $\tau$: one simulates the limiting stopping-time distribut…

Thompson-sampling conjecture. A Thompson-sampling-based algorithm may also work for the contextual dueling bandit setting because its stochastic nature encourages the agent to expl…

Let $a\in[K]\backslash\{a^*_0\}$ and $d\in[K]\backslash\{a^*_0,a\}$. Let $L^a_T$ denote the loss for arm $a$ at time horizon $T$, and let the semiparametric efficient score functio…

Let RMLP-2 be a contextual dynamic pricing method that assumes a log-concave noise distribution, and consider the simulation settings with standard Gaussian noise $F=\Phi(0,1)$ in…

Regret lower-bound conjecture. The obtained regret upper bound is close to the lower bound for this setting. The problem is harder than standard linear bandits and dynamic pricing…