Category: Evolutionary Intelligence

Artificial immune optimization system solving constrained omni-optimization

Abstract  This work investigates an artificial immune optimization system suitable for single and multi-objective constrained optimization.
In this optimizer, an evaluation index, which can decide the importance of individual in the current …

Abstract  

This work investigates an artificial immune optimization system suitable for single and multi-objective constrained optimization.
In this optimizer, an evaluation index, which can decide the importance of individual in the current population, is developed
to accelerate population division; the niching-like proliferation scheme is introduced to strengthen the diversity of population.
Thereafter, those diverse antibodies, with the help of immune evolution operations, evolve their structures along different
directions. Theoretical results show that such optimization system is convergent with low computational complexity. Experimentally,
one such optimizer is sufficiently examined by a suite of single and multi-objective test problems. Comparative experiments
illustrate that the optimizer with some striking characteristics is a potentially alternative optimization tool for constrained
omni-optimization.

  • Content Type Journal Article
  • Category Research Paper
  • Pages 1-16
  • DOI 10.1007/s12065-011-0064-1
  • Authors
    • Zhuhong Zhang, Institute of System Science and Information Technology, College of Science, Guizhou University, Guiyang, 550025 Guizhou, People’s Republic of China

Evolutionary computation as an artificial attacker: generating evasion attacks for detector vulnerability testing

Abstract  Intrusion detection systems protect our infrastructures by monitoring for signs of intrusions. However, intrusion detection
systems are themselves susceptible to vulnerabilities, which the attackers take advantage of to evade detec…

Abstract  

Intrusion detection systems protect our infrastructures by monitoring for signs of intrusions. However, intrusion detection
systems are themselves susceptible to vulnerabilities, which the attackers take advantage of to evade detection. In particular,
we focus on evasion attacks in which the attacker aims to generate a stealthy attack that eliminates or minimizes the likelihood
of detection. Attackers achieve stealth by mimicking normal behaviour while achieving the attack goals, hence bypassing the
detector. Previous work focused on generating evasion attacks using the internal knowledge of the detectors, hence adopting
a ‘white-box’ access to the detector. On the other hand, we adopt a ‘black-box’ approach and propose an evolutionary attacker
based on Genetic Programming. The access of our ‘black-box’ approach is limited to the feedback of the detector such as anomaly
rates and delays. We compare our ‘black-box’ approach with various ‘white-box’ approaches to investigate its effectiveness.
In doing so, the impact of anomalies from the break-in stage of the attacks and the delays based on locality frame counts
are also discussed. This is particularly important if the performance comparison is to reflect the real capabilities of detectors.

  • Content Type Journal Article
  • Category Research Paper
  • Pages 1-24
  • DOI 10.1007/s12065-011-0065-0
  • Authors
    • Hilmi Güneş Kayacık, School of Computer Science, Carleton University, 1125 Colonel By Drive, Ottawa, ON K1S 5B6, Canada
    • A. Nur Zincir-Heywood, Faculty of Computer Science, Dalhousie University, 6050 University Avenue, Halifax, NS B3H 1W5, Canada
    • Malcolm I. Heywood, Faculty of Computer Science, Dalhousie University, 6050 University Avenue, Halifax, NS B3H 1W5, Canada

Foreword: Special issue on Alan Turing

Foreword: Special issue on Alan Turing
Content Type Journal ArticleCategory EditorialPages 1-2DOI 10.1007/s12065-011-0063-2Authors
Christof Teuscher, ECE Department, Portland State University, Portland, OR, USA

Journal Evolutionary Intelligen…

Foreword: Special issue on Alan Turing

  • Content Type Journal Article
  • Category Editorial
  • Pages 1-2
  • DOI 10.1007/s12065-011-0063-2
  • Authors
    • Christof Teuscher, ECE Department, Portland State University, Portland, OR, USA

Evolving A-type artificial neural networks

Abstract  We investigate Turing’s notion of an A-type artificial neural network. We study a refinement of Turing’s original idea, motivated
by work of Teuscher, Bull, Preen and Copeland. Our A-types can process binary data by accepting a…

Abstract  

We investigate Turing’s notion of an A-type artificial neural network. We study a refinement of Turing’s original idea, motivated
by work of Teuscher, Bull, Preen and Copeland. Our A-types can process binary data by accepting and outputting sequences of
binary vectors; hence we can associate a function to an A-type, and we say the A-type represents the function. There are two modes of data processing: clamped and sequential. We describe an evolutionary algorithm, involving
graph-theoretic manipulations of A-types, which searches for A-types representing a given function. The algorithm uses both
mutation and crossover operators. We implemented the algorithm and applied it to three benchmark tasks. We found that the
algorithm performed much better than a random search. For two out of the three tasks, the algorithm with crossover performed
better than a mutation-only version.

  • Content Type Journal Article
  • Category Special Issue
  • Pages 1-20
  • DOI 10.1007/s12065-011-0062-3
  • Authors
    • Ewan Orr, Department of Physics and Astronomy, University of Canterbury, Christchurch, New Zealand
    • Ben Martin, Department of Mathematics and Statistics, University of Canterbury, Private Bag 4800, Christchurch, 8140 New Zealand

Using genetical and cultural search to design unorganised machines

Abstract  In 1948 Turing presented a general representation scheme by which to achieve artificial intelligence—his unorganised machines.
Significantly, these were a form of discrete dynamical system and yet dynamical representations remain…

Abstract  

In 1948 Turing presented a general representation scheme by which to achieve artificial intelligence—his unorganised machines.
Significantly, these were a form of discrete dynamical system and yet dynamical representations remain almost unexplored within
evolutionary computation. Further, at the same time as also suggesting that natural evolution may provide inspiration for
search mechanisms to design machines, he noted that mechanisms inspired by the social aspects of learning may prove useful.
This paper presents results from an investigation into using Turing’s dynamical representation designed by Evolutionary Programming
and a new imitation-based, i.e., cultural, approach. Moreover, the original synchronous and an asynchronous form of unorganised
machines are considered.

  • Content Type Journal Article
  • Category Special Issue
  • Pages 1-11
  • DOI 10.1007/s12065-011-0061-4
  • Authors
    • Larry Bull, Department of Computer Science, University of the West of England, Bristol, BS16 1QY UK

Alan Turing’s unorganized machines and artificial neural networks: his remarkable early work and future possibilities

Abstract  In a technical report submitted in 1948, Alan Turing presented a far-sighted survey of the prospect of constructing machines
capable of intelligent behaviour. This report was all the more remarkable for having been written at a tim…

Abstract  

In a technical report submitted in 1948, Alan Turing presented a far-sighted survey of the prospect of constructing machines
capable of intelligent behaviour. This report was all the more remarkable for having been written at a time when the first
programmable digital computers were just beginning to be built, leaving Turing with only paper and pencil with which to explore
his modern computational ideas. Turing may have been the first to suggest using randomly connected networks of neuron-like
nodes to perform computation, and proposed the construction of large, brain-like networks of such neurons capable of being
trained as one would teach a child. Some modern work has been performed on Turing’s neural networks, but all of it involves
the invention of new structures outside the scope of Turing’s original specifications in order to solve a technical error
on Turing’s part. I propose an alternative solution to Turing’s technical error, one which does not require the invention
of new structures, and outline an approach which may allow the better exploration of the particular properties of Turing’s
networks in their own right. I also give examples of ways in which to avoid viewing Turing’s early and strikingly original
work through the lens of the conventions of modern neural network theory. Using a “genetical” search to configure such networks,
as Turing suggested, is likely to yield non-intuitive or algorithmically opaque results, but this is likely to be a property
of brain-like networks themselves, and a sign we are approaching Turing’s initial goal.

  • Content Type Journal Article
  • Category Special Issue
  • Pages 1-9
  • DOI 10.1007/s12065-011-0060-5
  • Authors
    • Craig S. Webster, Centre for Medical and Health Sciences Education, Faculty of Medical and Health Sciences, University of Auckland, Private Bag 92-019, Auckland, New Zealand

Parameter control of metaheuristics with genetic fuzzy systems

Abstract  This paper introduces a genetic fuzzy system for parameter control of metaheuristics. Two basic metaheuristics have been considered
as examples, genetic algorithm and tabu search. The controlled parameters of the tabu search are th…

Abstract  

This paper introduces a genetic fuzzy system for parameter control of metaheuristics. Two basic metaheuristics have been considered
as examples, genetic algorithm and tabu search. The controlled parameters of the tabu search are the short and long term memories.
Parameters of the genetic algorithm under control are the mutation and reproduction rates. Fuzzy rule-based models offer a
natural mechanism to describe global behavior as a combination of control rules. They also inherit a means to gradually shift
between control rules which jointly defines a control strategy. They are a natural candidate to construct parameter control
strategies because they provide a way to develop decision mechanisms based on the specific nature of search regions and transitions
between their boundaries. An application example using the classic vehicle routing problem with time windows is included to
evaluate the genetic fuzzy system performance. Experimental results show that GFS-controlled metaheuristics improve search
behavior and solution quality when compared against standard, constant parameters genetic and tabu search approaches. It also
provides reasonably good suboptimal solutions faster than specially tailored exact methods reported in the literature.

  • Content Type Journal Article
  • Category Research Paper
  • Pages 183-202
  • DOI 10.1007/s12065-011-0059-y
  • Authors
    • Vitor Marques, School of Electrical and Computer Engineering, FEEC, University of Campinas, Unicamp, Campinas, SP, Brazil
    • Fernando Gomide, School of Electrical and Computer Engineering, FEEC, University of Campinas, Unicamp, Campinas, SP, Brazil

Advances in artificial immune systems

Advances in artificial immune systems
Content Type Journal ArticlePages 67-68DOI 10.1007/s12065-011-0058-zAuthors
Emma Hart, Institute for Informatics and Digital Innovation, Edinburgh Napier University, Edinburgh, UKChris McEwan, I2D3 Integrative I…

Advances in artificial immune systems

  • Content Type Journal Article
  • Pages 67-68
  • DOI 10.1007/s12065-011-0058-z
  • Authors
    • Emma Hart, Institute for Informatics and Digital Innovation, Edinburgh Napier University, Edinburgh, UK
    • Chris McEwan, I2D3 Integrative Immunology: Differentiation, Diversity, Dynamics, UPMC Univ Paris 06, Paris, France
    • Jon Timmis, Department of Computer Science and Department of Electronics, University of York, York, UK
    • Andrew Hone, School of Mathematics, Statistics & Actuarial Science, University of Kent, Kent, UK

Evolving scalable and modular adaptive networks with Developmental Symbolic Encoding

Abstract  Evolutionary neural networks, or neuroevolution, appear to be a promising way to build versatile adaptive systems, combining
evolution and learning. One of the most challenging problems of neuroevolution is finding a scalable and r…

Abstract  

Evolutionary neural networks, or neuroevolution, appear to be a promising way to build versatile adaptive systems, combining
evolution and learning. One of the most challenging problems of neuroevolution is finding a scalable and robust genetic representation,
which would allow to effectively grow increasingly complex networks for increasingly complex tasks. In this paper we propose
a novel developmental encoding for networks, featuring scalability, modularity, regularity and hierarchy. The encoding allows
to represent structural regularities of networks and build them from encapsulated and possibly reused subnetworks. These capabilities
are demonstrated on several test problems. In particular for parity and symmetry problems we evolve solutions, which are fully
general with respect to the number of inputs. We also evolve scalable and modular weightless recurrent networks capable of
autonomous learning in a simple generic classification task. The encoding is very flexible and we demonstrate this by evolving
networks capable of learning via neuromodulation. Finally, we evolve modular solutions to the retina problem, for which another
well known neuroevolution method—HyperNEAT—was previously shown to fail. The proposed encoding outperformed HyperNEAT and
Cellular Encoding also in another experiment, in which certain connectivity patterns must be discovered between layers. Therefore
we conclude the proposed encoding is an interesting and competitive approach to evolve networks.

  • Content Type Journal Article
  • Category Research Paper
  • Pages 145-163
  • DOI 10.1007/s12065-011-0057-0
  • Authors
    • Marcin Suchorzewski, Artificial Intelligence Laboratory, West Pomeranian University of Technology, ul. Żołnierska 49, 71-210 Szczecin, Poland

Co-evolution of lexical and syntactic classifiers during a language game

Abstract  This paper demonstrates for the first time that a learning classifier system can act as the core of a cognitive architecture
to allow agents to co-evolve lexical and syntactic conventions for the efficient communication of conceptu…

Abstract  

This paper demonstrates for the first time that a learning classifier system can act as the core of a cognitive architecture
to allow agents to co-evolve lexical and syntactic conventions for the efficient communication of conceptual strings during
a language game. The use of the learning classifier system has several inherent advantages to previous ad-hoc approaches to
modeling language games. It permits incremental addition of new populations of classifiers (e.g. syntactic classifiers) to
modify communicative conventions as necessary without destroying or catastrophically interfering with lower level (e.g. lexical)
conventions. The tendency for learning classifier systems to produce maximally general classifiers automatically produces
systematic syntactic conventions. These results provide a proof in principle that replication of classifiers in the brain
could serve an important role in cognition as proposed previously by the neuronal replicator hypothesis.

  • Content Type Journal Article
  • Category Research Paper
  • Pages 165-182
  • DOI 10.1007/s12065-011-0055-2
  • Authors
    • Chrisantha Fernando, Department of Informatics, University of Sussex, Falmer, Brighton, BN1 9RH UK