Challenges and opportunities in dynamic optimisation

Dynamic optimisation has been studied for many years within the evolutionary computation community. Many strategies have been proposed to tackle the challenge, e.g., memory schemes, multiple populations, random immigrants, restart schemes, etc. This talk will first review a few of such strate- gies in dealing with dynamic optimisation. Then some less researched areas are discussed, including dynamic constrained optimisation, dynamic combinatorial optimisation, time-linkage problems, and theoretical analyses in dynamic optimisation. A couple of theoretical results, which were rather unex- pected at the first sight, will be mentioned. Finally, a few future research directions are highlighted. In particular, po- tential links between dynamic optimisation and online learn- ing are pointed out as an interesting and promising research direction in combining evolutionary computation with ma- chine learning.