Workshops and Tutorials

Workshop

Title: 1st Australasian Workshop on Artificial Life and Computational Intelligence in Games

Organisers:

  • Fabio Zambetta (RMIT University)
  • Philip Hingston and Martin Masek (Edith Cowan University)
  • Kevin Wong (Murdoch University)

The first Australasian Workshop on AL & CI in Games to strengthen the local (Australian, with an eye to growing to the whole Asia-Pacific area) community on CI in computer games and build bridges between researchers in this space.

This event preceded ACALCI 2015 and was co-located at the University of Newcastle Conservatorium of Music, Newcastle, Australia.

For this workshop you did not need a paper. The day included an overview of current topics in ALife and Computational Intelligence in Games, and participants joined small teams and worked up ideas for collaborative projects.

Rather than presentation of papers, the workshop too a less conventional approach to a workshop, being more akin to a research "jam-session" i.e., a full day of intensive discussion on what each researcher has been (recently) working on and what collaborations might be built and on which topics.


Tutorials

Tutorial 1

  • Title: JIDT: An information-theoretic toolkit for studying the dynamics of complex systems
  • Presenter: Dr Joseph T. Lizier

Complex systems are increasingly being viewed as distributed information processing systems, particularly in the domains of Artificial Life, computational neuroscience and bioinformatics. This trend has resulted in a strong uptake in the use of information-theoretic measures to analyse the dynamics of complex systems in these fields. We introduce the Java information dynamics toolkit (JIDT): a Google code project which provides a standalone, (GNU GPL v3) open-source code implementation of information-theoretic measures of distributed computation in complex systems. The toolkit focusses on implementing measures for information dynamics, i.e. quantifying information storage, transfer and modification, and the dynamics of these operations in space and time. Principally, the toolkit provides measures for transfer entropy, (conditional) mutual information and active information storage, for both discrete and continuous-valued data.

Various types of estimator (e.g. Gaussian, Kraskov-Stoegbauer-Grassberger) are provided for each measure, and can be swapped at run-time due to Java's object-oriented polymorphism. Furthermore, while written in Java, the toolkit can be used directly in Matlab, Octave and Python. In this tutorial, we will introduce these information-theoretic measures of dynamics, present the JIDT distribution and discuss the design principles behind it, and walk-through several demonstrations to guide users on applying it on their own data sets.

Dr. Joseph Lizier is currently a Research Scientist at CSIRO in Sydney, Australia (since 2012). Previously he was a Postdoctoral Researcher at the Max Planck Institute for Mathematics in the Sciences& in Leipzig, Germany (2010-12), and a Senior Research Engineer at Telstra Research Laboratories (2001-06) and Seeker Wireless (2006-10) in Sydney. He received a PhD in Computer Science (2010), and Bachelor degrees in Engineering (2001, with University Medal) and Science (1999), all from the University of Sydney.

His research interests include information-theoretic approaches to complex systems, complex networks, artificial life and computational neuroscience. Joseph has received best paper awards at the IEEE Symposium on Artificial Life (2011 and 2013) and the Robocup Symposium (2013), is the 2013 CSIRO ICT Centre Young Scientist of the Year, and an Associate Editor of Frontiers in Robotics and AI.


Tutorial 2

  • Title: Evolving Neural Networks
  • Presenter: Dr David Howard

Nural networks are powerful nonlinear information processing models that are implicitly amenable to evolutionary search. When combined, neural network evolution (or neuro-evo) methods have found tremendous success as black-box controllers, time series predictors, and model approximators, to name but a few applications. This tutorial will begin by covering the basics - what is an evolutionary algorithm? How does evolution act on networks? What are the strengths and weaknesses of the approach?

Various types of neural network (Perceptrons, recurrent networks (RNN and CTRNN), and spiking networks) will be introduced, and a number of evolutionary approaches to network design will be practically demonstrated. Following a detour through a brief history of neuro-evolution applications, we will arrive at the state of the art in the field. Modern neuro-evo features, such as niching, plasticity, and deep learning nets will be discussed. Finally, numerous interesting "unanswered questions" will be presented.

David Howard received the MSc in Cognitive Systems from the University of Leeds, UK and PhD in Neuroevolutionary Learning Systems from the University of the West of England, UK. His interests are evolutionary algorithms , bio-inspired robotics, adaptive learning, and neural networks. In 2013 he won a prestigious CSIRO OCE fellowship, and moved to Brisbane to investigate the application of evolutionary algorithms to control, navigation, and guidance of quad rotor UAV platforms.


Tutorial 3

  • Title: Musical Metacreation
  • Presenters: Dr Oliver Bown, Dr Andrew Brown and Dr Toby Gifford

As regular music software adopts generative, autonomous and interactive capabilities, and an increasing number of practitioners involve metacreative processes in their work, the emerging field of "musical metacreation" (MuMe) has matured from an exploratory field to a more focused and applied one in which the practical application of adaptive and complex system is a core issue. This tutorial will develop the core themes of musical metacreation in light of contemporary issues of system design from the perspective of creative practitioner, commercial developer, computer scientist and interacting musician. A particular focus will be on emerging applications of MuMe technology.

The introductory tutorial will cover a range of MuMe topics that will introduce artificial life and computational intelligence researchers and students to current issues in the field, and lead on to a structured discussion of current MuMe research questions, particularly those that exhibit a strong overlap with ACALCI themes. The tutorial presenters are experienced creators of metacreative systems, and performances using their systems, and those of other practitioners, will be used as a discussion point in the tutorial. The tutorial will also cover issues of interaction design, evaluation and classification of systems, and the application of artificial life and computational intelligence techniques to musical autonomy.

Oliver Bown is postdoctoral research fellow at the Design Lab at the University of Sydney. He is a practicing electronic musician and a researcher working in creative technologies and computational creativity. He is a member of the organising committee of the Musical Metacreation Research Network.

Andrew R. Brown is Professor of Digital Arts at Griffith University in Brisbane, Australia. He is an active computer musician and computational artist. His research interests include digital creativity, computational aesthetics, music education and the philosophy of technology.

Toby Gifford is a music technologist, sound designer and acoustic musician. He has recently completed his PhD in interactive music systems. He is an active acoustic musician, live electronic music performer, and works at the arts/science nexus. One of his arts-science collaborations has been awarded the prestigous ANAT Synapse residency program for 2014. He currently lectures in music technology, interactive installation and sound design at the Queensland Conservatorium of Music.


For details about the tutorials please contact the tutorial chair Oliver Obst.