Reproduce “When Optimization Meets Human Intent: Human-Centered Diffusion Model Optimization” 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
Renye Yan, Jikang Cheng
The authors survey how human preferences are translated into computational proxies and used to optimize diffusion image generation. They organize work around intent representation, model-level updates, generation-level steering, trade-offs, and evaluation. The review identifies a potential Human–Proxy Gap: improved proxy scores do not establish better alignment with users’ intentions, and the synthesis does not show that every method fails users.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.