2-5 February 2016

Canberra, Australia


Yew-Soon Ong

Yew-Soon Ong Biography: Yew-Soon Ong is a Professor of Computer Science at the School of Computer Engineering, Nanyang Technological University, Singapore. He is Director of the A*Star SIMTECH-NTU Joint Lab on Complex Systems and Programme Principal Investigator of the Rolls-Royce@NTU Corporate Lab on Data Analytics and Complex Systems. He received a PhD degree on Artificial Intelligence in complex design from the Computational Engineering and Design Center, University of Southampton, United Kingdom in 2003. His current research interest in computational intelligence spans across memetic & evolutionary computation, machine learning, Big Data Analytics, and intelligent multi-agents. He is the founding Technical Editor-in-Chief of the Memetic Computing Journal, founding Chief Editor of the Springer book series on studies in adaptation, learning, and optimization, Associate Editor the IEEE Transactions on Evolutionary Computation, the IEEE Transactions on Neural Networks & Learning Systems, IEEE Computational Intelligence Magazine, IEEE Transactions on Cybernetics, IEEE Transactions on Big Data, and others. He has coauthored over 120 refereed publications and his research work on Memetic Algorithm was featured by Thomson Scientific's Essential Science Indicators as one of the most cited emerging area of research in August 2007. Recently, he received the 2015 IEEE Computational Intelligence Magazine Outstanding Paper Award and the 2012 IEEE Transactions on Evolutionary Computation Outstanding Paper Award for his work pertaining to Memetic Computing. Presently, he is Conference Chair of the Congress on Evolutionary Computation, World Congress on Computational Intelligence, Vancouver, Canada, 2016 and is secretary of the IEEE Transactions on Computational Intelligence and AI in Games steering committee.

Title: Multifactorial Optimization

Abstract:The design of population-based search algorithms of evolutionary computation (EC) has traditionally been focused on efficiently solving a single optimization task at a time. It is only very recently that a new paradigm in EC, namely, multifactorial optimization (MFO), has been introduced to explore the potential of evolutionary multitasking. The nomenclature signifies a multitasking search involving multiple optimization tasks at once, with each task contributing a unique factor influencing the evolution of a single population of individuals. MFO is found to leverage the scope for implicit genetic transfer offered by the population in a simple and elegant manner, thereby opening doors to a plethora of new research opportunities in EC, dealing, in particular, with the exploitation of underlying synergies between seemingly unrelated tasks. A strong practical motivation for the paradigm is derived from the rapidly expanding popularity of cloud computing (CC) services. It is noted that CC characteristically provides an environment in which multiple jobs can be received from multiple users at the same time. Thus, assuming each job to correspond to some kind of optimization task, as may be the case in a cloud-based on-demand optimization service, the CC environment is expected to lend itself nicely to the unique features of MFO.

In this talk, the formalization of the concept of MFO is first introduced. A fitness landscape-based approach towards understanding what is truly meant by there being underlying synergies (or what we term as genetic complementarities) between optimization tasks is then discussed. Accordingly, a synergy metric capable of quantifying the complementarity, which shall later be shown to act as a “qualitative” predictor on the success of multitasking. With the above in mind, a novel evolutionary algorithm (EA) for MFO is proposed, one that is inspired by bio-cultural models of multifactorial inheritance, so as to best harness the genetic complementarity between tasks. The salient feature of the algorithm is that it incorporates a unified solution representation scheme which, to a large extent, unites the fields of continuous and discrete optimization. The efficacy of the proposed algorithm, and the concept of MFO in general, shall finally be substantiated via a variety of computation experiments in intra and inter-domain evolutionary multitasking.