Model Inversion Networks for Model-Based Optimization

TL;DR

This paper introduces Model Inversion Networks (MINs), a novel approach for high-dimensional, data-driven optimization problems that effectively utilize offline data to find inputs maximizing unknown score functions.

Contribution

MINs provide a scalable, versatile method for inverse mapping from scores to inputs, addressing challenges of out-of-distribution inputs in high-dimensional spaces.

Findings

MINs outperform existing methods in high-dimensional optimization tasks.

MINs effectively leverage offline logged data for both contextual and non-contextual problems.

MINs demonstrate strong results in protein design, image optimization, and bandit problems.

Abstract

In this work, we aim to solve data-driven optimization problems, where the goal is to find an input that maximizes an unknown score function given access to a dataset of inputs with corresponding scores. When the inputs are high-dimensional and valid inputs constitute a small subset of this space (e.g., valid protein sequences or valid natural images), such model-based optimization problems become exceptionally difficult, since the optimizer must avoid out-of-distribution and invalid inputs. We propose to address such problem with model inversion networks (MINs), which learn an inverse mapping from scores to inputs. MINs can scale to high-dimensional input spaces and leverage offline logged data for both contextual and non-contextual optimization problems. MINs can also handle both purely offline data sources and active data collection. We evaluate MINs on tasks from the Bayesian optimization literature, high-dimensional model-based optimization problems over images and protein designs, and contextual bandit optimization from logged data.