Accelerating Inverse Learning via Intelligent Localization with Exploratory Sampling

TL;DR

This paper introduces iPage, a novel method that accelerates inverse learning by combining probabilistic inference from invertible models with deterministic gradient optimization and space-filling sampling, improving solution localization and exploration.

Contribution

The paper presents iPage, a new approach that integrates probabilistic inference and deterministic optimization to enhance inverse problem solving in materials and drug discovery.

Findings

iPage outperforms state-of-the-art methods on benchmark tasks.

The method effectively localizes solutions and explores parameter spaces.

Applications include quantum chemistry and additive manufacturing.

Abstract

In the scope of "AI for Science", solving inverse problems is a longstanding challenge in materials and drug discovery, where the goal is to determine the hidden structures given a set of desirable properties. Deep generative models are recently proposed to solve inverse problems, but these currently use expensive forward operators and struggle in precisely localizing the exact solutions and fully exploring the parameter spaces without missing solutions. In this work, we propose a novel approach (called iPage) to accelerate the inverse learning process by leveraging probabilistic inference from deep invertible models and deterministic optimization via fast gradient descent. Given a target property, the learned invertible model provides a posterior over the parameter space; we identify these posterior samples as an intelligent prior initialization which enables us to narrow down the search space. We then perform gradient descent to calibrate the inverse solutions within a local region. Meanwhile, a space-filling sampling is imposed on the latent space to better explore and capture all possible solutions. We evaluate our approach on three benchmark tasks and two created datasets with real-world applications from quantum chemistry and additive manufacturing, and find our method achieves superior performance compared to several state-of-the-art baseline methods. The iPage code is available at https://github.com/jxzhangjhu/MatDesINNe.