MLLGMay 23, 2019

Bayesian Optimization with Approximate Set Kernels

arXiv:1905.09780v312 citations
Originality Incremental advance
AI Analysis

This work addresses a domain-specific problem in optimization for machine learning, offering incremental improvements in efficiency for set-based tasks.

The paper tackled the problem of Bayesian optimization over permutation-invariant set inputs by developing a method with an approximate set kernel and constrained acquisition function optimization, resulting in outperformance over other methods in numerical experiments.

We propose a practical Bayesian optimization method over sets, to minimize a black-box function that takes a set as a single input. Because set inputs are permutation-invariant, traditional Gaussian process-based Bayesian optimization strategies which assume vector inputs can fall short. To address this, we develop a Bayesian optimization method with \emph{set kernel} that is used to build surrogate functions. This kernel accumulates similarity over set elements to enforce permutation-invariance, but this comes at a greater computational cost. To reduce this burden, we propose two key components: (i) a more efficient approximate set kernel which is still positive-definite and is an unbiased estimator of the true set kernel with upper-bounded variance in terms of the number of subsamples, (ii) a constrained acquisition function optimization over sets, which uses symmetry of the feasible region that defines a set input. Finally, we present several numerical experiments which demonstrate that our method outperforms other methods.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes