Computational Intelligence - February 2017 - 42

Review
Article

Kamran Shafi and Hussein A. Abbass
School of Engineering & Information Technology,
University of New South Wales, Canberra, AUSTRALIA

A Survey of Learning Classifier Systems in Games

Abstract

ames are becoming increasingly indispensable,
not only for fun but also
to support tasks that are more serious, such as education, strategic
planning, and understanding of
complex phenomena. Computational intelligence-based methods
are contributing significantly to
this development. Learning Classifier Systems (LCS) is a pioneering computational intelligence
approach that combines machine
learning methods with evolutionary
computation, to learn problem solutions in the for m of interpretable
rules. These systems offer several advantages for game applications, including a powerful and flexible agent
architecture built on a knowledgebased symbolic modeling engine;
modeling flexibility that allows integrating domain knowledge and different machine lear ning mechanisms
under a single computational framework; an ability to adapt to diverse
game requirements; and an ability to
learn and generate creative agent behaviors in real-time dynamic environments. We present a comprehensive
and dedicated survey of LCS in computer games. The survey highlights
the versatility and advantages of these
Digital Object Identifier 10.1109/MCI.2016.2627670
Date of publication: 6 January 2017

computer games industry is now
one of the most lucrative businesses
in the world with annual revenues
in tens of billions of dollars [8]. This
huge financial incentive is further
fueling research and development
in this domain.
Computational Intelligence (CI)
based [9] approaches provide a powerful set of methods and tools for
modeling artificial game-playing
agents and making games more
image licensed by ingram publishing
dynamic and interesting [10], [11].
systems by reviewing their application
The use of CI techniques in games has
in a variety of games. The survey is
grown steadily over the years because of
organized according to a general
advancements in related research areas,
game classification and provides an
such as evolutionary computation [12] and
opportunity to bring this important
deep neural networks [13]. A useful review
research direction into the public eye.
of CI techniques in games can be found
We discuss the strengths and weakin [14], [15].
nesses of the existing approaches and
Learning Classifier Systems (LCS) [16]
provide insights into important future
are one of the earliest CI approaches.
research directions.
LCS bring important opportunities to
games. First, LCS are rule-based, enabling
easier integration with the classic knowlI. Introduction
edge bases that are used in games. SecThe use of computer games is wideond, they are equipped with learning
spread in human life. Computer games
mechanisms that enable them to adapt
play diverse roles in shaping human life
the rules "on the fly." Third, recent
and knowledge, from being used for fun
advances in LCS introduce architectures
and entertainment [1] to being applied to
and algorithms that make them a techstrategic planning and decision support
nique that is fast and suitable for real[2], [3], health [4], defense [5], education
time environments.
and learning [6], and analysis of complex
LCS were introduced as biologically
phenomena [7]. In terms of revenue, the
inspired computational models for
human-like cognition. Typical LCS
Corresponding Author: Kamran Shafi (k.shafi@unsw
implementations consist of a knowledge
.edu.au; kamran.shafi@gmail.com)

IEEE ComputatIonal IntEllIgEnCE magazInE | FEbruary 2017

http://www.edu.au

Table of Contents for the Digital Edition of Computational Intelligence - February 2017