Abstract: Bayesian optimization (BO), a data-efficient method for expensive black-box optimization, has traditionally focused on single-task scenarios, ignoring potential correlations among related ...

This project implements a physics-informed neural network (PINN) that acts as a surrogate model for predicting key outcomes of the LPBF process (residual stress, porosity, geometric accuracy) based on ...

Autocomp is a portable, extensible framework for LLM-driven kernel optimization across tensor accelerators. Point it at a kernel, pick your hardware target, and Autocomp speeds it up, automatically.

Abstract: This article proposes a constrained evolutionary Bayesian optimization (CEBO) algorithm to cope with expensive constrained optimization problems with inequality constraints. The uniqueness ...