Unifying Likelihood-free Inference with Black-box Optimization and Beyond

TL;DR

This paper unifies likelihood-free inference and black-box optimization into a single probabilistic framework for biological sequence design, enabling new algorithms and improved performance in sequence optimization tasks.

Contribution

It introduces a unified probabilistic framework that connects likelihood-free inference with black-box optimization, and develops new algorithms for biological sequence design.

Findings

Previous optimization methods can be reformulated within the framework

Proposed algorithms outperform existing approaches in sequence design tasks

Framework facilitates flexible construction of sequence design methods

Abstract

Black-box optimization formulations for biological sequence design have drawn recent attention due to their promising potential impact on the pharmaceutical industry. In this work, we propose to unify two seemingly distinct worlds: likelihood-free inference and black-box optimization, under one probabilistic framework. In tandem, we provide a recipe for constructing various sequence design methods based on this framework. We show how previous optimization approaches can be "reinvented" in our framework, and further propose new probabilistic black-box optimization algorithms. Extensive experiments on sequence design application illustrate the benefits of the proposed methodology.