Title: Multi-fidelity EBB problems

Speaker: Nico Andres Thio

Summary: In the field of Multi-fidelity Expensive Black-Box (Mf-EBB), a problem instance consists of an expensive yet accurate source of information, and one or more cheap yet less accurate sources of information. The field aims to provide techniques either to accurately explain how decisions affect design outcome, or to find the best decisions to optimise design outcomes. Many techniques which use surrogate models have been developed to provide solutions to both aims. Only in recent years, however, have researchers begun to explore the conditions under which these new techniques are reliable, often focusing on problems with a single low-fidelity function, known as Bi-fidelity Expensive Black-Box (Bf-EBB) problems.

In this talk I will present an introduction to Bf-EBB problems. I will give the formulation of three variations of Bf-EBB problems and illustrate them with a toy problem. I will also discuss the question of how much techniques should rely on the low fidelity source, if at all, when approaching these problems.

Bio: Nico’s project focuses on the field of Multi-fidelity Expensive Black-Box problems. This field encompases any industrial problem where assessing the performance of a product design can be expensive, such as when building a prototype of a battery or plane, but other sources of information of this performance are available, which are both cheaper and less accurate.  The field aims to provide techniques either to accurately explain how decisions affect design outcome, or to find the best decisions to optimise design outcomes. Nico’s previous work analysed algorithms which focus on constructing networks (such as roads connecting cities) which are resilient to disaster failure, as well as creating models to simulate the behaviour of pedestrians under panic during evacuations in emergency scenarios.

WED 8 MARCH 9:30-10:30 (AEDT, Melbourne Time)

ZOOM MEETING ID: 873 1557 5255; PASSWORD: 778635