Implicitly Guided Design with PropEn: Match your Data to Follow the Gradient

TL;DR

PropEn is a novel framework that uses implicit guidance through sample matching to optimize models efficiently in data-scarce scientific domains, avoiding the need for large datasets and discriminator training.

Contribution

It introduces PropEn, a domain-agnostic, matching-based generative framework that approximates gradients for property optimization without requiring extensive data or discriminator models.

Findings

PropEn effectively guides design optimization in scientific applications.

It outperforms baseline methods in toy and real-world tasks.

Protein design results are validated with wet lab experiments.

Abstract

Across scientific domains, generating new models or optimizing existing ones while meeting specific criteria is crucial. Traditional machine learning frameworks for guided design use a generative model and a surrogate model (discriminator), requiring large datasets. However, real-world scientific applications often have limited data and complex landscapes, making data-hungry models inefficient or impractical. We propose a new framework, PropEn, inspired by ``matching'', which enables implicit guidance without training a discriminator. By matching each sample with a similar one that has a better property value, we create a larger training dataset that inherently indicates the direction of improvement. Matching, combined with an encoder-decoder architecture, forms a domain-agnostic generative framework for property enhancement. We show that training with a matched dataset approximates the gradient of the property of interest while remaining within the data distribution, allowing efficient design optimization. Extensive evaluations in toy problems and scientific applications, such as therapeutic protein design and airfoil optimization, demonstrate PropEn's advantages over common baselines. Notably, the protein design results are validated with wet lab experiments, confirming the competitiveness and effectiveness of our approach.