Interview with Dr. Joseph Lizier of University of Sydney, chair of the 2017 IEEE Symposium on Artificial Life (Interviewed by Dr. David Fogel, co-chair, 2017 IEEE Symposium Series on Computational Intelligence).
Dr. Lizier is an ARC DECRA Research Fellow and Senior Lecturer in Complex Systems in the Centre for Complex Systems and Faculty of Engineering and IT at the University of Sydney. He studies the dynamics of information processing in biological and bio-inspired complex systems and networks, particularly neural systems, and is teaching into The University of Sydney’s new Master of Complex Systems degree. Dr. Lizier is chairman of the upcoming 2017 IEEE Symposium on Artificial Life (ALIFE), to be held as part of the 2017 IEEE Symposium Series on Computational Intelligence, held in Honolulu, HI, Nov. 27-Dec. 1. I recently interviewed Dr. Lizier to learn about the latest artificial life research.
DF: Artificial life might be said to have had its start in the 1950s-1970s — I’m thinking for example of work by Nils Barricelli and also Michael Conrad — but when artificial life came into more prominence in the late 1980s there was a focus on studying “life as it could be, rather than life as it is.” In your view is that still the paradigm, or has it changed over time? If it’s changed, how so?
JL: Yes, “Life as it could be” rather than “Life as we know it” is still the binding call of the ALife community, which focusses on the simulation and synthesis of life-like systems. This means we study systems with life-like properties, that don’t necessarily occur in known biology. There are very good reasons for doing so, for example, in helping to understand natural systems in terms of constraints on the dynamics that can occur and why they occur, as well as with regards to designing artificial systems with life-like properties such as guided swarming behaviour. But let me add that while the paradigm is still the same, our science is moving fast, both in terms of making progress on traditional focus areas and in “expanding the tent” of areas that we focus on.
DF: What specific aspects of artificial life are you working on now?
JL: My research focuses on how biological and bio-inspired systems process information, using information theory to study how information is stored, transferred, and modified across the parts of these systems (known as information dynamics). There are several lines of work here, from very theoretical questions around, say, how to informationally decompose the effects of multiple sources on a target, through to more application-focussed investigations in computational neuroscience regarding characterising information flows revealed by brain imaging and how they relate to cognitive tasks. Several of these aspects are strongly embedded in the artificial life domain. For example, I’m trying to understand how the structure of a complex network gives rise to its function by using these measures of information processing which can be applied to any type of dynamics (from gene regulation to neural spiking activity). We’re also examining information flows between individuals partaking in collective swarm behaviour, so that we can pinpoint when, where and how collective decisions are made, in order to more completely describe the self-organisation taking place and/or identify any leadership behaviours.
DF: What do you hope to learn from this research?
JL: For me, the end-goal is a combination of both understanding (science) and design (engineering) of living and life-like complex adaptive systems. Since our measures of information processing can be studied on any type of dynamics, I think they offer us a cross-domain, directly comparable way to understand complex adaptive systems, how constraints led them to be the way they are and what they could be. For example, small-world networks are an example of a complex network pattern that occurs in an enormous range of biological (e.g. neural, genetic, social) and artificial (e.g. power distribution networks, Internet) systems. We know how to build them and why they’re easy to build, but what is it that they *do* that makes them so useful and promotes their persistence? How is that tied to their development? These questions can only be answered with a cross-domain approach like information dynamics. Answering these questions will put us in a more informed position in terms of designing life-like systems.
DF: What are others doing in artificial life research that you find particularly interesting?
JL: The most exciting aspects of artificial life are those of design approaches to life-like systems, particularly with regards to guided self-organisation. We know that self-organisation is a key property of life-like systems, offering advantages of scalability, flexibility to goal change and adaptability/robustness to damage — but how can we guide this process to achieve engineering goals? Are there intrinsic motivations or goals we can use to adapt/evolve life-like systems to position them well for general purpose use, with later refinement to achieve engineering goals? As an engineer, the question “How can we use this knowledge?” is always driving the types of scientific investigations I focus on, even though I’m mostly not explicitly working on design. Indeed, this is a good characterisation of what differentiates IEEE ALife from other artificial life conferences — being embedded inside the IEEE Symposium Series on Computational Intelligence (IEEE SSCI), it naturally has a stronger engineering focus.
DF: How have advancements in hardware (computing, robotics) helped change the ALIFE field over time?
JL: Deeply! These advances have enabled us to simply do more, and investigate more widely, deeply, and rigorously. We can examine more complicated systems because of robotics hardware, which can physically do more things now, such as achieving distributed self-assembly of complex structures. This has led to educational robots (for example, the e-puck) with many features, which have enabled a lot of research on social robotics. The combination of computing and robotic hardware itself is interesting, with robots that can run online simulations of their own behaviour before choosing their actions. Computing power has also enabled more intensive and greater numbers of simulations; an interesting example here is that of (IEEE ALife Organising Committee member) Mikhail Prokopenko and his team using high-performance computing clusters to explore fitness landscapes in designing tactics for their Robocup entry, which went on to win the World Championship in 2016. Importantly too, more computing power also introduces more rigor to our work, since we can perform longer studies to draw conclusions on the statistical significance of our findings.
DF: What would someone not working in ALife gain by attending your symposium?
JL: A friend who is a gastroenterologist said to me last week: “What I like about your research is that, while your applications are out of my area, I can always understand the broad principles and I can always relate it to a problem that I’m interested in.” So for starters, I guarantee you’ll be interested in what you see: Living and life-like systems exhibit self-organisation and emergence, which are some of the most fascinating properties to scientists and engineers. But beyond that, because ALife is considering such broad principles of life-like systems, it considers a wide range of application areas and incorporates many techniques, from complex networks to information theory, multi-agent systems to evolutionary computation — so you are highly likely to learn something that you can use in your own work. For example, the same approach we take to studying information processing in life-like systems can be used in neural analysis, pattern recognition and data mining, feature selection, financial market analysis, and so on. So please come and learn and share!
Joseph Lizier: The University of Sydney, Sydney, Australia email@example.com. http://lizier.me/joseph/
David Fogel: Natural Selection, Inc., 6480 Weathers Pl., Suite 350, San Diego, CA 92121, firstname.lastname@example.org. (858) 455-6449. www.natural-selection.com
The 2017 IEEE Symposium Series on Computational Intelligence can be found at: http://www.ieee-ssci.org
To read more:
See the hyperlinks contained in the interview, and also:
James M. Keller, Derong Liu, and David B. Fogel, Fundamentals of Computational Intelligence: Neural Networks, Fuzzy Logic, and Evolutionary Computation, John Wiley, NY, 2016
© 2017, David Fogel.