Life Sciences and Health Arthur Guijt, Seminar Anton Bouter

Failure modes of Linkage Learning and how the right neighborhood can solve them; Optimal Mixing Evolutionary Algorithms for Large-Scale Real-Valued Optimization
  • What Evolutionary Intelligence English
  • When 30-11-2021 from 16:00 to 17:00 (Europe/Amsterdam / UTC100)
  • Where online
  • Contact Name
  • Web Visit external website
  • Add event to calendar iCal


Join Zoom Meeting
https://cwi-nl.zoom.us/j/81535249101?pwd=ZDh3Y1kyZmxBUml1Tm00Rkx0NXBwdz09

Meeting ID: 815 3524 9101
Passcode: 967201

Title:      Failure modes of Linkage Learning and how the right neighborhood can solve them
Speaker:    Arthur Guijt
Abstract:   Linkage Learning is at the core of most of GOMEA's scalability improvements, and is a core component of the GOMEA family of approaches. Yet if Linkage Learning fails to work, performance is reduced drastically. In this presentation I'll briefly discuss the current approach, the implicit assumptions that are made, and the failure modes that may occur in practice. Furthermore, I will talk about using neighborhoods - more specifically Local Neighborhoods, how learning linkage locally resolves some of these common failure modes, and how (much) the definition of locality matters.

Title:      Optimal Mixing Evolutionary Algorithms for Large-Scale Real-Valued Optimization
Speaker:    Anton Bouter
Abstract:   It is known that to achieve efficient scalability of an Evolutionary Algorithm (EA), dependencies (also known as linkage) must be properly taken into account during variation. In a Gray-Box Optimization (GBO) setting, exploiting prior knowledge regarding these dependencies can greatly benefit optimization. We specifically consider the setting where partial evaluations are possible, meaning that the partial modification of a solution can be efficiently evaluated. Such problems are potentially very difficult, e.g., non-separable, multi-modal, and multi-objective. The Gene-pool Optimal Mixing Evolutionary Algorithm (GOMEA) can effectively exploit partial evaluations, leading to a substantial improvement in performance and scalability. Such partial evaluations have been applied to various real-world problems, including the optimization of treatment plans for prostate cancer, and deformable image registration.