Jiajun He

Discrete neural samplers achieve fixed-point convergence and drastically reduce mode collapse by leveraging Gibbs sampling within a masked diffusion framework.

METALICA, which implements Metadynamics on a pretrained diffusion model via Replica Exchange, is introduced, which implements Metadynamics on a pretrained diffusion model via Replica Exchange and accumulates a bias potential along a Collective Variable, repels new samples from previous ones through biased sampling, and reweights samples onto the unbiased distribution.

By rethinking hypothesis generation as a sampling problem, this framework boosts both the quality and diversity of scientific hypotheses, challenging the status quo of optimization-focused methods.