§½Ωεw VLSIZ^[ uο
2026N420ϊij 10:45`12:20
§½Ωεw@Ρν±E³ΒLpX@[LOΩ2K@WΟfUC[
QΑ³ΏEO\έsvEΞΚΜέ
Hyper-Heuristics:
Automated Design of Heuristics with Genetic Programming
Marko Đurasević
University of Zagreb
http://www.zemris.fer.hr/~idurasevic/
|
Abstract: |
|
|
Hyper-heuristic methods allow for
the automated design of novel heuristics for various combinatorial optimisation problems, thus relieving human experts of
this tedious task while also enabling the discovery of strategies that may
not be intuitive or are difficult to design manually. By operating at a
higher level of abstraction, these methods can generate heuristics based on
the characteristics of the problem instance, leading to more robust and
general-purpose solution approaches. Among the various hyper-heuristic
methods, genetic programming, a method based on evolutionary computing, is
most commonly used in practice due to its potential of generating
interpretable heuristics. This talk will provide an introduction into the
topic of hyper-heuristics with the focus being placed on genetic programming,
and will showcase its application to several selected combinatorial optimisation problems. |
|
|
Speaker
Biography: |
|
|
Marko Đurasević |
Marko Đurasević
received B.Sc., M.Sc. degrees in software engineering and his Ph.D. in
computer science from the University of Zagreb, Faculty of Electrical
Engineering and Computing (FER), Zagreb, Croatia, in 2012, 2014, and 2018,
respectively. From October 2014 he is working at the Department of
Electronics, Microelectronics, Computer and Intelligent Systems at University
of Zagreb, Faculty of Electrical Engineering and Computing as a teaching and
research assistant. In December 2019 he was promoted to Assistant Professor. Currently
he is the principal investigator of the research project gDesign and
Enhancement of Domain Specific Languages (DEDSL)h, international project with
industry, AVL-AST d.o.o. for Advanced Simulation Technologies Development,
seated in Zagreb, and AVL Graz - Instrumentation and Test Systems. He is also
currently participating in the project gHyperheuristic
Design of Dispatching Rules (HyDDRa)h, financed by
the Croatian Science Foundation. His research interests include genetic
programming, scheduling, operations research, soft computing, and
hyper-heuristics. |
|
βΉζ: yRGViHw@dqξρHwΘj |
|
