Download Flyer: UGA Clemson 2026 Flyer_0.pdf (136.78 KB)
Title: Are We There Yet? A Probabilistic Journey to Global Optima
Abstract: Global optimization over non-convex landscapes is notoriously difficult—akin to a long road trip with no map, no gradients, and lots of local distractions. Traditional algorithms, including many metaheuristics, often struggle with high dimensionality, sensitivity to initial conditions, or inaccessibility of derivative information—frequently converging to suboptimal solutions (if at all). In this talk, I will introduce ProGO—a Probabilistic Global Optimizer designed to answer that age-old question: “Are we there yet?”—without needing gradients, perfect guesses, or a magic wand. Built on a sequence of multidimensional integrations, ProGO is backed by a solid convergence theory and mild assumptions (no deal-breakers here).The heart of our method lies in a latent slice sampler that zips toward nascent optima distributions with geometric speed—no stalling, no detours. Unlike many existing methods, ProGO scales gracefully with dimension and consistently finds the global optima across a wide range of challenging test functions. In empirical comparisons, it leaves gradient-based and many gradient-free optimizers in the dust—especially in terms of regret and convergence speed. One caveat: if your function takes a week to compute, we suggest packing snacks—ProGO’s strength is in smart sampling, not expensive function evaluations.
Event Type
- NISS Sponsored