Reproduce “SPREAD: Sampling-based Pareto front Refinement via Efficient Adaptive Diffusion” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Sedjro Salomon Hotegni, Sebastian Peitz
SPREAD uses a Transformer-based diffusion model to search for sets of designs that balance competing objectives. It refines candidate batches through objective-informed gradient guidance and radial-basis-function repulsion, improving trade-offs while spreading points across the Pareto front. On engineering benchmark RE21, its online hypervolume was 70.10 ± 0.01 versus 48.14 ± 0.00 for PMGDA, but results vary across tasks and settings.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.