Learning Low-Dimensional Embeddings for Black-Box Optimization
By: Riccardo Busetto , Manas Mejari , Marco Forgione and more
Potential Business Impact:
Finds good answers faster in tricky problems.
When gradient-based methods are impractical, black-box optimization (BBO) provides a valuable alternative. However, BBO often struggles with high-dimensional problems and limited trial budgets. In this work, we propose a novel approach based on meta-learning to pre-compute a reduced-dimensional manifold where optimal points lie for a specific class of optimization problems. When optimizing a new problem instance sampled from the class, black-box optimization is carried out in the reduced-dimensional space, effectively reducing the effort required for finding near-optimal solutions.
Similar Papers
Towards Universal Offline Black-Box Optimization via Learning Language Model Embeddings
Machine Learning (CS)
Lets computers solve many different problems.
Instance Generation for Meta-Black-Box Optimization through Latent Space Reverse Engineering
Machine Learning (CS)
Makes computer problem-solvers learn better on new tasks.
Clustering-based Meta Bayesian Optimization with Theoretical Guarantee
Machine Learning (CS)
Finds best settings faster, even with many past tries.