283. Dimensionality reduction techniques for derivative free optimization
Invited abstract in session MB-1: Advances in Large-Scale Derivative-Free Optimization , stream Zeroth and first-order optimization methods.
Monday, 10:30-12:30Room: B100/1001
Authors (first author is the speaker)
| 1. | Coralia Cartis
|
| Mathematical Institute, University of Oxford |
Abstract
We will discuss linear, and also, time permitting, nonlinear techniques for reducing the dimension of problems and algorithms in the absence of derivatives. We will discuss various aspects of subspace techniques in model-based optimization problems, both randomised and deterministic, first- and second-order, low-rank and full rank. Numerical experiments and theoretical results will be presented. Time permitting, we will present machine learning based dimensionality reduction techniques that we use in the context of Bayesian optimization methods.
Keywords
- Black-box optimization
- Derivative-free optimization
Status: accepted
Back to the list of papers