Neural Genetic Search in Discrete Spaces

TL;DR

Neural Genetic Search (NGS) is a novel test-time search method that integrates genetic algorithm principles into deep generative models, enhancing their performance across diverse tasks.

Contribution

This paper introduces NGS, a versatile and easy-to-implement search algorithm that combines genetic algorithms with deep generative models for improved test-time performance.

Findings

Effective in routing problems

Generates adversarial prompts for language models

Facilitates molecular design

Abstract

Effective search methods are crucial for improving the performance of deep generative models at test time. In this paper, we introduce a novel test-time search method, Neural Genetic Search (NGS), which incorporates the evolutionary mechanism of genetic algorithms into the generation procedure of deep models. The core idea behind NGS is its crossover, which is defined as parent-conditioned generation using trained generative models. This approach offers a versatile and easy-to-implement search algorithm for deep generative models. We demonstrate the effectiveness and flexibility of NGS through experiments across three distinct domains: routing problems, adversarial prompt generation for language models, and molecular design.