Goal: minimize unknown function $\phi$ in as few evaluations as possible

• $\phi$: drawn randomly from the prior
• $c(x_t)$: cost of getting new data point, expected budget constraint

Algorithm:

1. Build posterior $f \given y$ using data $(x_1,\phi(x_1)),..,(x_t,\phi(x_t))$
2. Find optimum of acquisition function $\alpha_{f\given y}$ and evaluate $\phi$ at $$ \htmlClass{fragment}{ x_{t+1} = \argmax_{x\in\c{X}} \alpha_{f\given y}(x) } $$

Optimal choice $x_{t+1}$ and when to stop: intractable dynamic program

Cost-aware baseline: expected improvement per unit cost $$ \begin{aligned} \htmlClass{fragment}{ \alpha_{t}^{\f{EIPC}}(x) } & \htmlClass{fragment}{ = \frac{\f{EI}_{f\given y_1,..,y_t}(x; \max_{1\leq\tau\leq t} y_\tau)}{c(x)} } & \htmlClass{fragment}{ \f{EI}_\psi(x; y) } & \htmlClass{fragment}{ = \E \max(0, \psi(x) - y) } \end{aligned} $$

Should we open the closed box? Maybe!

Depends on costs $c$, reward distribution $f$, and value of open box $g$

One closed vs. open box: Markov decision process

Optimal policy: open if $\f{EI}_f(x; g) \gt c(x)$

Consider how optimal policy changes as a function of $g$

If both opening and not opening is optimal: $g$ is a fair price

Define: $\alpha^\star_t(x) = g$ where $g$ solves $\f{EI}_f(x; g) = c(x)$

Theorem (Weitzman, 1979). Let:

• $X$ be a finite set,
• $f : X \to \R$ be a finite-mean random function for which $f(x)$ is independent of $f(x')$ for $x \neq x'$,
• $c : X \to \R_+$, without loss of generality, be deterministic.

Then, for the cost-per-sample problem, the policy defined by maximizing the Gittins index acquisition function $\alpha^\star$ with its associated stopping rule is Bayesian-optimal.

Gittins index $\alpha^\star$: optimal for cost-per-sample problem

• What about expected budget-constrained problem?

Theorem. Assume the expected budget constraint is feasible and active. Then there exists a $\lambda \gt 0$ and a tie-breaking rule such that the policy defined by maximizing the Gittins index acquisition function $\alpha^\star(\.)$, defined using costs $\lambda c(x)$, is Bayesian-optimal.

Proof idea: Lagrangian duality

Our work: extends special case of a result of Aminian et al. (2024) to non-discrete reward distributions

Bayesian optimization: posterior distribution is correlated

Define Pandora's Box Gittins Index acquisition function:

$\alpha^{\f{PBGI}}(x) = g$ where $g$ solves $\f{EI}_{f\given y}(x; g) = c(x)$

Correlations: incorporated into acquisition function via the posterior

$\lambda$: controls risk-averse vs. risk-seeking behavior

Limit as $\lambda\to0$: converges to UCB with automatic learning rate

Counterexample: random objective with high-variance high-cost region

Pandora's Box Gittins Index: still performs well

Novel acquisition function: Pandora's Box Gittins Index

• Settings: expected budget-constrained and cost-per-sample
• Works in heterogeneous-cost setting and for uniform costs
• Closely-related to both expected improvement and UCB
• Exact optimality in simplified problem: orthogonal insights

Performance of Pandora's Box Gittins Index

• Sufficiently-easy low-dim. problems: comparable to baselines
• Too-difficult high-dim. problems: similar to random search
• Medium-hard problems of moderate dim.: strong performance
• Can compete with state-of-the-art non-myopic approaches

Gittins Index Theory

• Workhorse tool in queuing theory
• Minimize expected wait time: serve short jobs first

Bayesian Optimization: high-dimension and complex feedback models

• Freeze-thaw
• Continuous-time and asynchronous
• Bayesian quadrature
• Function networks
• Exact optimality in simplified problems without dependence

Unexplored toolkit with which to understand decision-making