Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Bayesian Algorithm Execution: Estimating Computable Properties of Black-box Functions Using Mutual Information

About

In many real-world problems, we want to infer some property of an expensive black-box function $f$, given a budget of $T$ function evaluations. One example is budget constrained global optimization of $f$, for which Bayesian optimization is a popular method. Other properties of interest include local optima, level sets, integrals, or graph-structured information induced by $f$. Often, we can find an algorithm $\mathcal{A}$ to compute the desired property, but it may require far more than $T$ queries to execute. Given such an $\mathcal{A}$, and a prior distribution over $f$, we refer to the problem of inferring the output of $\mathcal{A}$ using $T$ evaluations as Bayesian Algorithm Execution (BAX). To tackle this problem, we present a procedure, InfoBAX, that sequentially chooses queries that maximize mutual information with respect to the algorithm's output. Applying this to Dijkstra's algorithm, for instance, we infer shortest paths in synthetic and real-world graphs with black-box edge costs. Using evolution strategies, we yield variants of Bayesian optimization that target local, rather than global, optima. On these problems, InfoBAX uses up to 500 times fewer queries to $f$ than required by the original algorithm. Our method is closely connected to other Bayesian optimal experimental design procedures such as entropy search methods and optimal sensor placement using Gaussian processes.

Willie Neiswanger, Ke Alexander Wang, Stefano Ermon• 2021

Related benchmarks

TaskDatasetResultRank
DiscoBAXTau Protein Assay (test)
Runtime (s)113.2
2
DiscoBAXInterferon-Gamma Assay (test)
Runtime (s)97.03
2
Level Set EstimationHimmelblau (test)
Runtime (s)14.97
2
Level Set EstimationVolcano (test)
Runtime (s)289.9
2
Local OptimizationHartmann 6D (test)
Runtime (s)7.64
2
Local OptimizationAckley 10D (test)
Runtime (s)29.31
2
Top-kRosenbrock (k = 6) (test)
Runtime (s)18.31
2
Top-kGB1 k = 10 (test)
Runtime (s)865.9
2
Showing 8 of 8 rows

Other info

Follow for update