Evolutionary One-Step Generators: Fast and Diverse Sampling for Discrete Design

Key points

• Several discrete design tasks, such as molecular discovery, require diverse collections of useful candidates at low computational cost.1
• High validity alone does not guarantee a useful candidate library: repeatedly generating the same valid structures leaves few distinct alternatives.1
• The method combines distribution matching with structural constraints and optional diversity or history-dependent rewards, using antithetic low-rank evolution strategies without requiring criterion-specific differentiable surrogates.1
• On molecular generation benchmarks, our compact generator achieves over $50\times$ the valid-and-unique yield per estimated dense operation compared to recent one-step flow-map baselines while retaining high chemical validity.1

Sources (1)

• [1]Evolutionary One-Step Generators: Fast and Diverse Sampling for Discrete Design
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 6, 01:54 PM

  To address this challenge, we propose EGO (Evolutionary Generators with One-step inference), a framework for training compact generators directly on discrete outputs.

  Several discrete design tasks, such as molecular discovery, require diverse collections of useful candidates at low computational cost.

Extractive summary: sentences quoted from the sources.