Clonal Selection Algorithms

JASON BROWNLEE
Technical Report 070209A
Complex Intelligent Systems Laboratory, Centre for Information Technology Research,
Faculty of Information Communication Technology, Swinburne University of Technology
Melbourne, Australia
jbrownlee@ict.swin.edu.au

Abstract-Inspired by Darwins theory of natural selection to In brief, the principle of the theory is that the antigen
explain the diversity and adaptability of life, Burnets clonal (the foreign molecule that the immune system is
selection theory explains the diversity and learning properties of defending against) selects those lymphocytes (B-cells or
the acquired immune system of vertebrates. In a similar white blood cells that detect and stop antigens) with
mirroring manner to the field of evolutionary computation that receptors capable of reacting with a part of the antigen.
attempts to use the principles of the Darwinian theory and Selection results in the rapid proliferation of the selected
genetics to address practical engineering problems, a new field of cell to combat the invasion (clonal expansion and
study called Clonal Selection Algorithms has emerged that production of antibodies). During this cell duplication
attempts the same task by abstracting and applying the process coping errors occur (somatic hypermutation)
principles of Burnets foundational immunological theory. This which may result in an improved affinity of the progeny
paper provides a summary of this new field of clonal selection cells receptors for the triggering antigen.
algorithms and proposes an algorithm taxonomy, a standardized
This description clearly sounds like a selective and
nomenclature, and a general model of such algorithms. Finally,
stochastic-based adaptive process. This general method
the field is compared and contrasted to the field of evolutionary
has inspired the field of clonal selection algorithms
computation, and general research trends are discussed.
which attempt to harness its potential in the application
Keywords- Clonal Selection Algorithm, CSA, Clonal Selection
to primarily optimization and classification problem
Theory, Clonal Selection Principle, Artificial Immune System, domains.
Algorithm Review, CLONALG, AIRS, BCA, MISA, IA The paper is broken down as follows; Section II
presents a general model of clonal selection algorithms
I. INTRODUCTION which includes a standard definition, specification of
Artificial Immune Systems (AIS) is the investigation algorithmic principles, and a standardized nomenclature.
of models and abstractions of the vertebrate (typically Also presented is a high-level algorithm taxonomy for
mammalian) immune system and the application of these interpreting the current state of the art in clonal selection
models and algorithms to practical endeavours such algorithms. Section III applies the taxonomy from
computation problem domains in the fields of science, section II.C and reviews the field of clonal selection
engineering, and information technology [88]. Although algorithms. This review is complete1 and focuses on
the source of inspiration for computational models in the applications and general algorithmic principles. Section
immune system is near limitless, four main sub fields of IV addresses the clear similarity between clonal
research have emerged in AIS cantered on prominent selection algorithms and some evolutionary algorithms,
immunological theories; negative selection algorithms contrasting the two related fields of research. Finally V
(NSA), immune network algorithms (INA), danger general trends observed from the literature review are
theory algorithms (DTA), and clonal selection discussed, and some potential future areas for clonal
algorithms (CSA). selection algorithm design are suggested.
Like Darwin s theory of theory of accumulated blind II.GENERAL MODEL
variation in the face of natural selection that
revolutionized the field of biology [19], Burnet s theory The clonal selection theory is the foundational
(inspired by Darwin s) of clonal selection suitably principle of modern immunology, thus it is tightly
explains the laboratory observations of antibody interconnected with other immunological theories. This
diversity in the acquired immune system and follows for the algorithms inspired by such theories. For
transformed modern immunology [43-45]. A discussion example, negative selection algorithms model
of Burnet s theory is beyond the scope of this work (for a classification problems in the complement space
modern introduction see [13,112,153]), although it is not although still rely on the clonal selection principle to
the theory that is relevant, but rather the computational iteratively improve exemplars. Immune network
principles that can be drawn. Thus, it is popular in the algorithms for clustering and optimization use the
field of clonal selection algorithms to describe the excitation and suppression properties of the network
inspiration as the clonal selection principle as opposed

model though also use the clonal selection principle for

to the clonal selection theory , as it is the principle that

is being applied rather than the theory that is being 1

To the authors knowledge at the date of publication and given the
investigated. scattered nature of publications in this field and intense googling

CIS Technical Report 070209A February 2007 page 1 of 13

iterative refinement of their models. A more complete 1. Randomly initialise pool of antibodies
review of the field would include a discussion of the 2. Expose the pool to antigen
clonal selection properties of other immunological a. Clonal Selection
algorithms such as immune network and negative
b. Clonal Expansion
c. Somatic Hypermutation
selection algorithms, and this remains an exercise for
Figure 1 - General algorithmic model of the clonal selection principle
future work.
Definition 1.0: A clonal selection algorithm is primarily Where the pool is exposed to 1 antigen, the
focused on mimicking the clonal selection principle which is
operator s selection and expansion are affinity
composed of the mechanisms; clonal selection, clonal
proportionate, and mutation is affinity inversely-
expansion, and affinity maturation via somatic hypermutation.
proportionate.

C.Taxonomy
A.Nomenclature
Before an algorithm taxonomy is presented, it is
Given that inspiration is such a critical feature to this useful to present a brief taxonomy of the broader field of
field of study, it is important to have a consistent research. As stated in the introduction the term chosen
nomenclature when describing clonal selection for the field of study is Clonal Selection Algorithms

algorithms. This nomenclature is drawn from the (CSA) inspired by the Clonal Selection Principle (CSP)

biological inspiration and refers to the principles which is derived from the Clonal Selection Theory

mimicked by inspired algorithms (see Table 1 for a (CST). The field belongs to the study of Artificial
listing of common CSA terms drawn from the literature). Immune Systems (AIS) which is commonly associated
General Clonal Selection Algorithms with Biologically-Inspired Computation (BIC) or
Candidate Solution, Exemplar Antibody, B-Cell, Lymphocyte Computational Intelligence (CI).
Collection of New Samples, Progeny Clone
Elitism, Memory Memory Set, Memory Cell A taxonomy of clonal selection algorithms has not
Generalization Cross-reactivity been presented before2, and although obvious is
Learning Principle Clonal Selection Principle expected to be useful in interpreting the current state of
Mutation, Variation Hypermutation the field. A algorithmic-genealogical approach was
Population, Collection of Samples Repertoire taken similar to that used by Galeano, Veloza-Suan, et
Re-sampling Principle, Improvement Affinity Maturation
al. [64] in the comparative analysis of artificial immune
Re-sampling, Reproduction Cloning, Clonal Expansion
Selection Antigen-Antibody Matching
network models. Here, the lineage is defined by the
Solution Quality, Fitness Affinity, Avidity seminal algorithm names, as follows; the Artificial
Immune Recognition System (AIRS), the B-Cell
Table 1 - Clonal Selection Algorithms Common Nomenclature
Algorithm (BCA), the Clonal Selection Algorithm
(CLONALG), the Immunological Algorithm family
B.Archetype Algorithm (IA), Multi-objective Immune System Algorithm
Cutello and Nicosia [144] suggest clonal selection (MISA), and Other for unclassified works (see Table 2).
algorithms take two key features into account; the Lineage Algorithms Primary
hypermutation and the clonal expansion mechanisms. Application
They go on to describe hypermutation as a local search AIRS AIRS, AIRS2, Parallel AIRS Classification
procedure that leads to fast maturation, and the clonal BCA BCA Optimization
expansion phase triggers growth of a new population of CLONALG CSA, CLONALG, CLONALG (1,2), Optimization
(CSA) ACS, CLONCLAS, RCSA, MOCSA,
useful B-cells focused on the triggering antigen. They IMCSA, AISMM, SACSA, ECA
also propose that the primary immune response may be IA (SIA) IA, SIA, I-PAES, CLIGA, CLIGA+, Optimization
taken as a training phase, whereas the improved NC-IA, READ-Alg, opt-IA, opt-
secondary response may be taken as the testing phase. IMMALG, Par-IA, Dyn-IMMALG
MISA MISA Multi-Objective
de Castro and Timmis [88] also suggest the two key Optimization
features of clonal selection algorithms are the mutation Other Too large to classify at this time Optimization
and cloning properties, and go on to outline more
Table 2 Basic algorithm genealogy
specific properties of these mechanisms ([88] page 80):
Principle 1.0: The proliferation rate of each immune cell is
proportional to its affinity with the selective antigen (higher the relative III.ALGORITHMS
affinity, the more progeny) This section presents a review of clonal selection
Principle 1.1: The mutation suffered by each immune cell during
algorithms applying the taxonomy presented in II.C.
reproduction is inversely proportional to the affinity of the cell receptor with
Sections A through to E present the five main algorithm
the antigen (higher the relative affinity, the lower the mutation)
lineages. Section F summarizes uncategorized works,
and section G summarizes those works claimed or
They also suggest that selection plays a critical role referred to be clonal selection algorithms which do not
in both the strong selective pressure during affinity meet definition 1.0.
maturation, and in the selection of long lived memory
cells.
Thus a general clonal selection algorithm possesses
the following mechanisms: 2
To the best knowledge of the authors at the time of writing.

CIS Technical Report 070209A February 2007 page 2 of 13

A.Clonal Selection Algorithm (CLONALG) P1 <- select(P, n) // clonal selection
Hidden at the back of a technical report on the
ForEach p1 of P1 Do // clonal expansion
C <- clone(p1)
applications of artificial immune systems de Castro and EndFor
Von Zuben [86] proposed the Clonal Selection ForEach c of C Do // affinity maturation
Algorithm (CSA) as a computational realization of the hypermutate(c)
clonal selection principle for pattern matching and EndFor
optimization. This algorithm which has become perhaps
ForEach c of C Do // presentation
affinity(c)
the most popular in the field of AIS, was later published EndFor
and represented [84], and again [85] where it was P <- insert(C, n) // greedy selection
renamed to CLONALG (CLONal selection ALGorithm). Pr <- rand(d, L)
P <- replace(P, d, Pr) // random replacement
The general CLONALG model involves the selection EndWhile
of antibodies (candidate solutions) based on affinity Figure 2 - CLONALG pseudocode listing
either by matching against an antigen pattern or via
evaluation of a pattern by a cost function. Selected In an attempt to exploit the inherent distributedness

antibodies are subjected to cloning proportional to of the immune system, Watkins, Bi, et al. [12] propose
affinity, and hypermutation of clones inversely- that each antibody in the algorithms repertoire can be
proportional to clone affinity. The resultant clonal-set treated independently given the lack of inter-antibody
competes with the antibody population for membership interactions. The pattern recognition variation of the
in the next generation, and finally low-affinity CLONALG was modified such that each memory cell is
population members are replaced by randomly generated partitioned to different processes and evolved
antibodies. The pattern recognition variation of the independently or in small groups, the results from which
algorithm includes a maintenance memory solution set are collated at the end of the algorithm run and returned
which in its entirety represents a solution. A binary- as the algorithm result3. White and Garret [54] also
encoding scheme is employed for the binary-pattern investigated the pattern recognition version of
recognition and continuous function optimization CLONALG and generalized the approach for the task of
examples, and an integer permutation scheme is binary pattern classification renaming it Clonal
employed for the Travelling Salesman Problem (TSP) Classification (CLONCLAS) where their approach was
example. compared to a number of simple Hamming distance
CLONALG Description and Pseudocode based heuristics.
Walker and Garrett [59] investigated CLONALG and
Evolution Strategies (ES) on dynamic function
optimization, showing that although CLONALG can
Parameter Description
P Repertoire of antibodies
N The fixed antibody repertoire size achieve better results faster than ES on low dimensional
n The number of antibodies to select for cloning dynamic functions, ES consistently outperforms
L Bit string length for each antibody CLONALG on the two high-dimensional problems
Nc Number of clones created by each selected antibody. tested. In an attempt to address concerns of algorithm
Originally expressed as a function of the repertoire size
efficiency, parameterization, and representation selection
(for optimization) N c = round ( N ) (where is a for continuous function optimization Garrett [122]
user parameter), although a direct integer specification proposes an updated version of CLONALG called
of Nc is simpler. A rank-based (affinity proportionate) Adaptive Clonal Selection (ACS). The mutation
variation of the question is presented for pattern
recognition. parameter, the number of antibodies selected for cloning,
d Number of random antibodies to insert at the end of and the number of clones produced for each antibody
each generation. Random antibodies replace the d lowest were changed to automatic parameters, controlled in a
affinity antibodies in the repertoire similar way to those in Evolution Strategies (ES).
Stop condition Typically a specified number of generations or function
evaluations. Cutello, Narzisi, e al. [141] proposed two modified
affinity Solution evaluation, typically the solution is decoded versions called CLONALG1 and CLONALG2 with
into a domain specific representation and assigned a varying elitist strategies which were raced against the
quality costing opt-IA algorithm. Dilettoso and Salerno [34] treated
clone Duplication of a bit string.
hypermutate Modification of a bit string where the flipping of a bit is
CLONALG as a niching technique and raced it against
governed by an affinity proportionate probability traditional EC niching approaches. Wang [157] proposed
distribution. Originally p = exp( f ) , although the a CSA based on CLONALG with a static clone sized
1
applied to power filter design observing niching like
opt-aiNET variant is also popular p = exp( f ) behaviours. Cruz-Cortes, Trejo-Perez, et al. [110]

investigated CLONALG with binary and gray encoding
(where is a user parameter and f is the normalized

schemes as well as a real-valued encoding scheme with a

affinity scoring). mutation scheme based on Gaussian and Cauchy random
Table 3 - CLONALG parameters numbers.
P <- rand(N, L)
While Not StopCondition Do 3
ForEach p of P Do // presentation The choice of application is poor, given that the binary pattern
affinity(p) recognition task was selected by de Castro and Von Zuben for demonstration
EndFor purposes only.

CIS Technical Report 070209A February 2007 page 3 of 13

 Babayigit, Akdagli, et al. [14] applied CLONALG to Network algorithm (AINE) to represent clones (groups)
locating good model parameters for the null synthesizing of identical B-cells. The AIRS is a clonal selection
of linear antenna arrays by amplitude control. Given inspired procedure of cloning and somatic
their reported success, the authors applied the algorithm hypermutation for preparing a set of real-valued
to other antenna design problems [2,81]. Campelo, exemplars suitable for classifying unobserved cases and
Guimaraes, et al. [36] proposed a Real-coded Clonal uses a single iteration over a set of training data.
Selection Algorithm (RCSA) with Gaussian-based Watkins and Boggess [11] quickly went on to apply the
mutation applied to the electromagnetic design AIRS to a suite of benchmark classification problems,
optimization problem (called the TEAM workshop and Goodman and Boggess problems [28] did the same,
problem 22). Campelo, Guimaraes, et al. [37] also comparing the approach to the similar Learning Vector
proposed a multi-objective version of their algorithm Quantization (LVQ) approach.
called MOCSA. This variation was later applied to the Given the rapid popularity of the approach Marwah
same electromagnetic design problem in [38]. Another and Boggess [46] investigated the algorithm seeking
multi-objective application of CLONALG was proposed issues that affect the algorithms performance. The
by Stevens, Das, et al. [22]. compared various variations of the algorithm with
Dong, Shi, et al. [149] proposed the Immune Memory modified resource allocation schemes, tie-handling
Clonal Selection Algorithm (IMCSA) applied to within the ARB pool and ARB pool organization. AIRS
designing stack filters for noise suppression. This was again raced againt LVQ by Boggess and Hamaker
extension to CLONALG used dual-binary strings in each [97] on datasets that contained irrelevant features to
antibody, self-tuning mutation parameters, assess the algorithms ability to handle noise. Greensmith
recombination parameters and inserted memory cells and Cayzer applied AIRS to hierarchal document
that were developed using alternative algorithms. Acan classification [66] which culminated in Greensmith s
[1], proposed an extension called Artificial Immune masters work [65].
System with Mutation Multiplicity (AISMM) that used Watkins and Timmis [8] proposed a new version of
multiple concurrent mutation operators in the application the algorithm called AIRS2 which became the
to continuous function optimization. Bian and Qiu [164] replacement for AIRS1. The updates reduced the
applied CLONALG to PMU placement, and Amaral, complexity of the approach while maintaining the
Amarak, et al. [63] applied CLONALG to parameter accuracy of the results. An investigation by Goodman,
tuning in PID controller design. Boggess, et al. [29] into the source of the AIRS so called
CLONALG has also been hybridized with many power indicating that perhaps the memory cell
other optimization procedures, some examples include maintenance procedures played an important role. The
the following: Zuo and Fan [168] proposed the Chaotic approach was compared to some state of the art
Search Immune Algorithm (CSIA) that integrated classification algorithms. A follow-up empirical
elements of the CLONALG algorithm and was applied investigation by Goodman and Boggess [27] supported
to the tuning Radial-Basis Functions (RBF) in real-time the original finding indicating that the process by which
controller design. Zhong, Zhang, et al. [171] proposed new memory cells are admitted into the ARB pool is
the Simulated Annealing Clonal Selection Algorithm critical to the success of the approach.
(SACSA) which hybridizing CLONALG with SA in the Using work on parallelizing the CLONALG [12] as
application to classification. This approach was extended a basis, Watkins and Timmis [9] proposed a parallel
by Zhong, Zhang, et al. [170] and renamed to the version of AIRS permitting the division of training
Unsupervised Artificial Immune Classifier (UAIC). patterns and memory pool suitable to exploit parallel
Wang, Wang, et al. [113] combined CLONALG with hardware. An empirical study of various non-Euclidean
Particle Swarm Optimization (PSO) and applied it to distance measures was performed by Hamaker and
function optimization. Karakasis and Stafylopatis [136] Boggess [53] assisting application to mixed-variable
combined CLONALG with Gene Expression classification domains. Finally a large study of both
Programming (GEP) called Enhanced Clonal Algorithm version of AIRS was published by Watkins, Timmis, et
(ECA) and applied the approach to data mining. al. [10], and culminated in Watkins dissertation [5].
Litvinenko, Bidyuk, et al. [135] created a similar hybrid
algorithm and applied the approach to time series Jin, Bie, et al. [166] extended AIRS and applied the
prediction. approach to software quality classification. Meing,
Putten, et al. [92] benchmarked AIRS determining that
B.Artificial Immune Recognition System (AIRS) the classifier is quite stable. Xu, Chow, et al. [82,83]
After CLONALG, the Artificial Immune Recognition applied AIRS to power outage cause identification with
System (AIRS) algorithm is perhaps the second most an imbalance of training cases. Finally, Polat, Shan, et
popular clonal selection algorithm, although the al. [78,79] extended AIRS to make use of fuzzy logic
approach was designed for and has only been applied to rules called FS-AIRS. The authors later applied the
the supervised classification problem domains. The approach to ECG data [80], and later renamed the
earliest work on AIRS was in Watkins masters thesis approach to Fuzzy-AIRS [77]. Garain, Chakroborty, et
[4], although was later published in [6]. The approach is al. domain [128] propose a CSA inspired by AIRS and
a supervised learning algorithm for classification that CLONALG for optical character recognition. This work
uses the idea of an Artificial Recognition Ball (ARB) was extended and applied to a more difficult multiple-
introduced in earlier works on the Artificial Immune class pattern recognition problem in characters [129].

CIS Technical Report 070209A February 2007 page 4 of 13

C.B-Cell Algorithm (BCA) Clark, Hone, et al. [35] using a Markov chain model.
Kelsey and Timmis [61] proposed the B-Cell The proof simplifies the algorithm to an elitist search
Algorithm (BCA) as an AIS designed for continuous with a single population member suggesting that the
function optimization. The algorithm maintains a pool of members of the population can be treated independently
B-cells (binary-encoded candidate solutions) that are given the lack of interaction during the optimization
subjected to cloning and mutation. An elitist replacement procedure. Further, they speculate that the introduction
population maintenance scheme is applied that ensures of inter-solution interactions in the BCA will have a
only improved cells are admitted into the pool. The detrimental effect on the number of function evaluations
mutation operator, which is called contiguous somatic

during a search. Finally in a recent empirical study Bull,

hypermutation selects a random sub-string of a solution Knowles, et al. [111] the authors apply the BCA to what
to probabilistically vary, what the authors claim as hot-

they refer to as less-smooth test problem instances

spot mutation. Kelsey, Timmis, et al. [60] applied the (Diophantine equations) seeking empirical convergence
BCA to multimodal-dynamic chaotic test functions. heuristics. Four variants of the algorithm are compared,
Empirical algorithm tuning by the authors revealed small an approach that uses an elitist selection mechanism to
population sizes (3-5, 12) show better results. introduce inter-solution interactions, and three of what
the authors refer to as megamutation schemes that

In an investigation of AIS applied to optimization, attempt to introduce further diversity into the search.
Hone and Kelsey [7] provide a case study investigation These less-greedy modifications of to the achieve better
of the BCA and show apparently fractal structures on the final results compared to the classical BCA on the test
complex plain suggesting the potential usefulness of functions chosen, perhaps suggesting the use of diversity
studying AIS as nonlinear dynamical systems. In a introduction approaches in further BCA applications.
further empirical study, Timmis, Edmonds, et al. [62]
compare the BCA to opt-aiNET4 and the HGA D.Immunological Algorithm Family (IA)5
attributing the partial success of BCA to the mutation
scheme, speculating it results in the escaping of local- A simple clonal selection inspired algorithm was
optima search behaviour. proposed by Cutello and Nicosia [142,143] called
Immunological Algorithm (IA) later renamed to Simple
BCA Description and Pseudocode Immunological Algorithm (SIA) [144]. The algorithm
maintains a population of B-cells that are exposed to a
clonal expansion process each generation. This
expansion process involves the cloning of cells and the
Parameter Description
P Repertoire of antibodies
N Antibody population size application of a hypermutation operator. The algorithm
Nc Number of clones to create of each antibody was demonstrated on the Minimum Hitting Set Problem
nR Number of random antibodies to create and insert each (MHSP) and the 3-Satisifiability Problem (3-Sat).
generation
Stop Condition Typically if no progress is made for a number of The SIA was extended and an applied to the Graph
generations Colouring Problem (GCP) [146]. The extensions
hypermutation Uses a processes called contiguous hypermutation, a involved the introduction of a local-search procedure
random location in the bit string is selected, and a that operated upon each B-cell after the clonal expansion
random bub-string length is selected. Each bit in the
phase. In addition, rather than an elitist selection method


substring is flipped with the probability

of maintain the population size after each expansion, an

replace A parent is replaced only if a member of its clone has a aging operator was introduced for the B-cells. B-cells
higher affinity (greedy replacement)
are probabilistic deleted from the population using an
Table 4 - BCA parameters equation inspired by the biological literature. Two
variations of the aging operator were applied, an elitist
P <- rand(N, L) version that ensured the best B-cell s were not deleted,
and a pure strategy that probabilistically deleted
While Not StopCondition Do
ForEach p of P Do // presentation
affinity(p) irrespective of the elitist concerns. A birthing operator
was also added to top-up the population to the

EndFor
ForEach p of P Do // clonal expansion configured level as needed, and an information gain (a
C <- clone(p, Nc)
stabilization in the measure of information discovered by
the algorithm) measure was used as the termination
C <- rand(nR, L) // random insertion
ForEach c of C Do // affinity maturation
hypermutate(c) criteria for the algorithm.
The probabilistic aging operator was replaced with a
EndFor
ForEach c of C Do // presentation
affinity(c) simplified generational aging operator by Cutello,
EndFor Nicosia, et al. [148] in an application to the 2DHP
C <- best(C) protein folding problem. In a more detailed study on
P <- replace(c, p) // clonal selection
different varieties of the same protein folding domain,
the aging operator was further tweaked to facilitate
EndFor
EndWhile
Figure 3 - BCA pseudocode listing
5
The Immunological Algorithm (IA) is renamed and represented many
A proof of convergence for the BCA is proposed by
times by its authors. Other names include Simple Immune Algorithm (SIA),
Cloning Information Gain Aging (CLIGA), and Optimization Immune
4
(opt-aiNET) an immune network algorithm as specified in [87] Algorithm (opt-IA, opt-IMMALG).

CIS Technical Report 070209A February 2007 page 5 of 13

longer life spans on some B-cells deemed useful to the EndWhile
search process here [134]. The clonal expansion aspect Figure 4 - SIA pseudocode listing
of the algorithm (cloning and hypermutation) was
grafted into to an existing evolutionary multiple-object An extension to opt-IA was proposed by Cutello,
optimization technique [140] and called I-PAES. Nicosia, et al. [132] called parallel immune algorithm
The transformed algorithm (generational aging, (Par-IA) which is a master-slave version of the algorithm
information gain stopping criteria) was reviewed and applied to numerical function optimization. Cutello,
renamed to the Cloning, Information Gain, Aging

Nicosia, et al. [133] again renamed the approach to

(CLIGA) algorithm [144]. A modified version called Optimization Immune Algorithm (Opt-IMMALG),
CLIGA+ was proposed in which each B-cell contains applying the approach to continuous function
more than one receptor (pattern), permitting application optimization using a real-valued representation as
of the algorithm to pattern recognition tasks. Also opposed to the binary representation used in previous
proposed in this work is a Noisy Channel variation of works. Also stated in this work is the use of fitness
SIA (NC-IA), and a Reaction-Diffusion variation of SIA inversely proportional hypermutation as the standard
(READI-Alg) both of where were applied to instances of mutation operator for the approach. This algorithm is
the GCP. extended and renamed dynamic immune algorithm (dyn-
IMMALG) by Cutello, Lee, et al. [130] who propose a
Cutello, Narzisi, et al. [139] again renamed the dynamic rather than static clonal operator. The approach
approach to Optimization Immunological Algorithm is applied to binary trap functions and compared to opt-
(opt-IA) and applied the approach to instances of binary IA and variations of CLONALG.
trap functions. An additional fitness inversely-
proportional hypermutation (referred to as E.Multi-objective Immune System Algorithm (MISA)
hypermacromutation ) schemes was proposed and

compared to the traditional static approach. This Coello Coello and Cruz Cortes [18] introduce an AIS
algorithm was evaluated again in a larger study called the Multi-objective Immune System Algorithm
involving a number of machine learning domains [141], (MISA), and as its name suggests was designed as a
and again on a large number of continuous function population-based approach for constrained and
optimization instances [131]. unconstrained multi-objective optimization. In the
approach a repertoire of solutions is split into antigens
Cutello, Nicosia, et al. [147] investigate the (Pareto non-dominated and feasible solutions) and
hypermutation operators of opt-IA. Cutello, Morelli, et antibodies (Pareto dominated and infeasible solutions).
al. [138] apply opt-IA to the 3DHP protein folding A bit-string representation is used and antigens are
problem. Cutello and Nicosia [145] apply opt-IA to selected at random and matched against antibodies
graph colouring, MHSP, and satisfiability. Work by (using Hamming distance). After selection, antibodies
Anile, Cutello, et al. [3] hybridizes opt-IA with a direct are cloned and mutated the population is unioned and
search method. An extension of opt-IA called aligner reduced back to the configured size, culling the lower
was proposed by Cutello, Lee, et al. [137] applied to quality solutions. An external (elitist) memory repertoire
multiple sequence alignment of DNA. is maintained of non-dominated feasible solutions.
SIA Description and Pseudocode Solutions are added to the memory set if they are non-
dominated by the current memory set population and
sufficiently diverse as determined by a grid-based
Parameter Description maintenance structure.
P Antibody population
l Length of binary string representation MISA was extended by the same authors [109] and
d Population (repertoire) size further compared to state of the art evolutionary
dup The number of clones created for each antibody approaches for multi-objective optimization. An EC-AIS
clone Duplication of the bit string hybrid terminology was adopted and an EC-based
hypermutation Probabilistic modification of a bit string (bit flipping), crossover mechanism was adopted within the memory
requires the specification ( ) of the probability of

set. The main algorithm was simplified such that all

flipping each bit. population members were consider antibodies, and only
Table 5 - SIA parameters the lower score solutions are selected for cloning and
hypermutation. The modified algorithm was shown
P <- rand(d, l) effective, although it demonstrated rapid convergence
ForEach p of P Do // presentation behaviours on benchmark problem instances. Finally,
Villalobos-Arias, Coello Coello, et al. [104] proposed a
affinity(p)
EndFor
While Not StopCondition Do convergence proof for the update MISA showing that the
ForEach p of P Do // clonal expansion elitism within the algorithm was needed to guarantee
Pc <- clone(p, dup) convergence.
EndFor
ForEach c of Pc Do // affinity maturation
hypermutate(c) F.Miscellaneous Works
EndFor This section lists works that do not neatly fit into the
ForEach c of Pc Do // presentation
above rough grouping of works. The works contained in
this section are not canonical, are new, or are less
affinity(c)
EndFor
P <- select(Pc, P, d) // clonal selection frequently referenced. It should be noted that the

CIS Technical Report 070209A February 2007 page 6 of 13

majority are recent (within the last two years), Bentley, et al. [68-74] have a body of work on a
application works (as opposed to new algorithms or Dynamic Clonal Selection (DynamiCS) which is a T-cell
theory), inspired or derived from CLONALG (with or inspired negative selection approach for intrusion
without reference) and produced by non-western detection that uses clonal selection mechanisms during
research groups (mainly from China). See Table 6 for a detection generation.
summary of these works arranged by general application
domain6. IV.RELATION TO EVOLUTIONARY COMPUTATION
General Application References Evolutionary Computation (EC) is a field of study
Feature selection for model [41,93,107,152,159-162] much like the field of AIS, although draws its inspiration
Parameter tuning for model or controller [26,39,55,95,105,127,163,167] for computation from Darwinian (theory of natural
Parameter tuning for a PID controller [30,31] selection) and neo-Darwinian (findings of modern
Anomaly and or intrusion detection [40,67,99,108] genetics also referred to as the new synthesis) theories of
Pattern recognition [52,89]
Multi-objective optimization [16,20,90,102,150,156]
evolution. The field of EC is also more established, and
[116,151] clonal selection algorithms bear a superficial similarity
Function optimization [32,47- to some EC algorithms such as the Genetic Algorithm
51,56,91,96,101,118,119,154,15 (GA), and properties of modern Evolution Strategies
5,172,175,176,178,180,181] (ES). See [23,24] for a classical treatment of EC and
General optimization [15,17,21,42,94,117,121,165,17 [124,125] for a modern treatment of the field of EC.
9]
Multi-user detection [57,57,98,100,103,173] In their work on CLONALG, de Castro and Von
Hybridized with other algorithm(s) [33,58,158,174] Zuben [86] address the similarity between a GA and
Table 6 - Summary of uncategorized application CAS works their approach, particularly in regard to the binary
representation used and the stochastic-Darwinian
processes employed by both. They go on to suggest the
G.Outliers
differences, include the vocabulary used (genetics and
This section pertains to algorithms and works which evolution verses the shape-space formalism and
superficially appear to be clonal selection algorithms, antibody-antigen cellular interactions), and the somatic
theoretical investigations into the principle commonly mutation and receptor editing used to explore the shape
sited as algorithms, and algorithms which have been space. The authors [85] later claim that CLONALG can
labelled as such although do not meet the definition be categorized as an evolutionary-like algorithm ,

outlined in section II. They are listed here for although they maintain the same arguments of
completeness, although by no means is this a complete inspiration, vocabulary, and formalism and the primary
collection of ambiguous clonal selection algorithm differences.
works.
In their book de Castro and Timmis [88] again
Forrest, Javornik, et al. [123] investigated the pattern acknowledge the similarity of CLONALG to an EA.
recognition properties of the immune system. They used They are quick to point out that a major difference
a binary coded genetic algorithm (GA) to model between the inspirations of the two approaches is that
antibody-antigen matching in the immune system, which mutation in evolution is random, whereas the
included the clonal selection mechanism, claiming The hypermutation process of clonal selection is controlled
GA without crossover is a reasonable model of clonal and directed proportional to the receptors affinity with
the triggering antigen. They go on to suggest that work


selection, while the GA with crossover models genetic

evolution . Hightower, Forrest, e al. [115] use a Binary on EA s can be leveraged by CSA s indicating that

GA model of somatic hypermutation of clonal selection research on selection operators (e.g. tournament, roulette
to investigate the Baldwin Effect and evolution. wheel, etc.) may be exploited. The shape-space
Weinand [114] propose a dynamical systems formalism is presented as a CSA representation
computational model of somatic mutation of B cells to abstraction and alternative representation schemes and
evaluate the affect of somatic mutation of affinity corresponding mutation mechanisms are discussed, also
maturation of the immune response. Fukuda, Mori, et al. leveraging from research from representation and
[126] and Mori, Tsukiyama, et al. [76] use a GA to mutation in EA s.
investigate clonal selection properties and immune
network algorithms for scheduling and resource de Castro and Timmis provide a treatment of
allocation. evolution and the clonal selection of the acquired
immune system. They suggest an important difference
Zhang and Hou propose a Niching Clonal Selection between the two theories is the fact that in the clonal
Algorithm (NCSA) [169] that combines negative process expansion occurs through cell cloning, that
selection and refinement using CSA applied to the there is no sex or genetic recombination, rather only
pattern matching problem of anomaly detection. Yo and affinity inversely-proportionate somatic hypermuatation.
Hou [177] proposed an extension of CLONALG called
CsAL (Clonal Selection Algorithm) to investigate the In work on the MISA Coello Coello and Cruz Cortes
negative selection approach to virus detection. Kim, [18] claim that their approach is not a genetic algorithm
because it does not use recombination (crossover
operator), they later [109] adopt a crossover procedure in
6
Many of these works could not be obtained by the author prior to the their approach, as well as adopt a hybrid EC and AIS
publication of this paper, thus general application domain was assumed in terminology.
some cases from paper abstracts

CIS Technical Report 070209A February 2007 page 7 of 13

 In their work evaluating the BCA on function family).
optimization Timmis, Edmonds, et al. [62] suggest that This work attempted to unify the field of clonal
BCA is not a GA based on the empirical performance of selection algorithms by presenting 1) a specific and
the approach on a small suite of test problem instances, reusable definition of clonal selection algorithms, 2) a
although they are very quick to point out the limitations general model for interpreting clonal selection
of their small study and the requirement for further algorithms, and 3) a general taxonomy for interpreting
research. the current state of clonal selection algorithms research.
The niching-like properties (an EA property inspired It appears from the literature review that clonal
by theories of population genetics and ecology) were selection algorithms are suitable for optimization
observed by de Castro and Von Zuben [85] with domains and for classification domains. It is important to
CLONALG on multimodal function optimization and note that CSA research is still in its infancy and these
empirically compared to the fitness sharing approach of applications may be considered merely demonstrations
Goldberg, et al. [24,25,75]. The niching search of capability of the general method.
properties are conjectured to occur given the hill-
climbing like behaviour of the independent and semi- Finally, from an algorithm design perspective, there
independent evolution of B-cells of the various clonal are likely many aspects of the clonal selection principle
selection algorithms. Thus, there may be a connection which have not been realized in the current state of the
between CSA and greedy stochastic-based hill climbing art clonal selection algorithms. A study of the
algorithms such as the real-valued hill-climber used by immunological theory from a biological rather than
Mahfoud [120] when evaluating niching EA s, and the algorithm vantage as well as associated physiology will
binary hill climbers used to evaluate early GA s [106]. likely reveal not only computationally interesting
mechanisms and architectures, but may also suggest
To summarize: suitable general application domains. Some plausible
1. The primary difference between EA s and CSA s areas for future clonal selection algorithm investigation
is the inspiration, resulting in differing may include the distributedness of the immune system
abstractions and nomenclature. physiology and the autonomy of the clonal response, the
2. The secondary difference between EA s and specific cellular and or genetic mechanisms employed
CSA s is the operators, the specific adaptive during clonal expansion, and physiology and
mechanisms employed by both approaches. mechanisms of the antibody-antigen binding during the
immune response.
3. The principles inspiring the algorithms are very
similar in terms of their general adaptive method, ACKNOWLEDGMENTS
specifically; selection, reproduction, and Tim Hendtlass his patience and for providing useful
variation. feedback on drafts of this paper
4. Some CSA s (CLONALG, BCA, IA family) may
be viewed as specialized EA s without crossover, REFERENCES
meaning some research on EA s is likely to be [1] A Acan, "Clonal selection algorithm with operator multiplicity,"
applicable to work on CSA s. CEC2004. Congress on Evolutionary Computation, CEC2004, pp. 1909-1915,
2004.
V. CONCLUSIONS [2] A. Akdagli, K. Guney, and B. Babayigit, Clonal Selection
Algorithm for Design of Reconfigurable Antenna Array with Discrete Phase
This review of clonal selection algorithms revealed at Shifters Journal of Electromagnetic Waves and Applications, vol. 21, pp.
least three interesting findings. 215-227, 2007.
[3] A.M. Anile, V. Cutello, G. Narzisi, G. Nicosia, and S. Spinella,
The first being the clear and recent popularity in the "Lipschitzian pattern search and Immunological Algorithm with quasi-
application of CLONALG derivatives, particularly from Newton method for the Protein Folding Problem: An innovative multistage
eastern (Chinese) works. The volume of works suggests approach," Proceedings of the Internation Workshop on Natural and Artificial
that potential ease of implementation of the approach Immune Systems, NAIS 2005, Italy, pp. 307-323, 2006.
and the variety of application (primary engineering [4] Andrew B. Watkins, AIRS: A resource limited artificial immune
optimization and model refinement) suggests the classifier 2001. Mississippi State University.
potential generality of the approach. The second [5] Andrew B. Watkins, Exploiting Immunological Metaphors in the
Development of Serial, Parallel, and Distributed Learning Algorithms 2005.
interesting trend is the hybridization of the algorithm University of Kent.
into other algorithms and systems. In addition to using [6] Andrew B. Watkins and Lois C. Boggess, "A Resource Limited
CSA s as the general iterative adaptive element in other Artificial Immune Classifier," Proceedings of Congress on Evolutionary
artificial immune system algorithms, the CSA principle Computation, HI, USA, pp. 926-931, May 2002.
has been grafted into a variety of other algorithms [7] Andrew Hone and Johnny Kelsey, "Optima, Extrema, and
including particle swarm optimization (PSO), gene Artificial Immune Systems," Proceedings Artificial Immune Systems: Third
expression programming (GEP), evolutionary algorithms International Conference, ICARIS 2004, Catania, Sicily, Italy, pp. 80-90,
2004.
and various other adaptive methods. The third interesting
[8] Andrew Watkins and Jon Timmis, "Artificial Immune Recognition
finding is the clear similarity of the clonal selection System (AIRS): Revisions and Refinements," 1st International Conference
principle with Darwin s evolution principle, and the on Artificial Immune Systems (ICARIS2002), University of Kent at
resultant strong similarity between evolutionary Canterbury, pp. 173-181, 2002.
algorithms (the genetic algorithm in particular) and some [9] Andrew Watkins and Jon Timmis, "Exploiting Parallelism Inherent
clonal selection algorithms (CLONALG, BCA, IA in AIRS, an Artificial Immune Classifier," Proceedings of the 3rd
International Conference on Artificial Immune Systems (ICARIS2004),

CIS Technical Report 070209A February 2007 page 8 of 13

Catania, Sicily, Italy, pp. 427-438, 2004. [30] Dong Hwa Kim, "Robust Intelligent Tuning of PID Controller for
[10] Andrew Watkins, Jon Timmis, and Lois Boggess, Artificial Multivarlable System Using Clonal Selection and Fuzzy Logic," Knowledge-
Immune Recognition System (AIRS): An Immune-Inspired Supervised Based Intelligent Information and Engineering Systems, pp. 848-853, 2005.
Learning Algorithm Genetic Programming and Evolvable Machines, vol. 5, [31] Dong Hwa Kim, Jae Hoon Jo, and H. Lee, "Robust power plant
pp. 291-317, Sep, 2004. control using clonal selection of immune algorithm based multiobjective,"
[11] Andrew Watkins and Lois Boggess, "A New Classifier Based on Proceedings of the Fourth International Conference on Hybrid Intelligent
Resource Limited Artificial Immune Systems," Proceedings of Congress on Systems, HIS 2004, pp. 450-455, 2004.
Evolutionary Computation, Honolulu, USA, pp. 1546-1551, May 2002. [32] Du Haifeng, Jiao Licheng, and Liu Ruochen, "Adaptive polyclonal
[12] Andrew Watkins, Xintong Bi, and Amit Phadke, "Parallelizing an programming algorithm with applications," Proceedings. Fifth International
Immune-Inspired Algorithm for Efficient Pattern Recognition," Intelligent Conference on Computational Intelligence and Multimedia Applications.
Engineering Systems through Artificial Neural Networks: Smart Engineering ICCIMA 2003, pp. 350-355 , 2003.
System Design: Neural Networks, pp. 225-230, 2003. [33] E. Carpaneto, G. Chicco, and M. Starda, "Applications of immune
[13] Arthur M. Silverstein, The Clonal Selection Theory: what it really and clonal selection-based techniques to distribution system optimal
is and why modern challenges are misplaced Nature Immunology, vol. 3, pp. operational planning," IEEE Mediterranean Electrotechnical Conference,
793-796, 2002. 2006. MELECON, pp. 1024-1027, 2006.
[14] B. Babayigit, A. Akdagli, and K. Guney, A clonal selection [34] E. Dilettoso and N. Salerno, A self-adaptive niching genetic
algorithm for null synthesizing of linear antenna arrays by amplitude control algorithm for multimodal optimization of electromagnetic devices IEEE
Journal of Electromagnetic Waves and Applications, vol. 20, pp. 1007-1020, Transactions on Magnetics, vol. 42, pp. 1203-1206, Apr, 2006.
Jun, 2006. [35] Edward Clark, Andrew Hone, and Jon Timmis, "A markov chain
[15] B.K. Panigrahia, Salik R. Yadavb, Shubham Agrawalb, and M.K. model of the B-cell algorithm," Artificial Immune Systems, pp. 318-330,
Tiwaric, A clonal algorithm to solve economic load dispatch Electric Power 2005.
Systems Research, vol. 2006. [36] F. Campelo, F. G. Guimaraes, H. Igarashi, and J. A. Ramirez, A
[16] Bin Lu, Licheng Jiao, Haifeng Du, and Maoguo Gong, "IFMOA: clonal selection algorithm for optimization in electromagnetics IEEE
Immune Forgetting Multiobjective Optimization Algorithm," Advances in Transactions on Magnetics, vol. 41, pp. 1736-1739 , May, 2005.
Natural Computation, First International Conference, ICNC 2005, Changsha, [37] F. Campelo, F.G. Guimar es, R.R. Saldanha, H. Igarashi , S.
China, pp. 399-408, 2005. Noguchi, D.A. Lowther, and J.A. Ramirez, "A novel multiobjective immune
[17] Caihong Mu and Mingming Zhu, "Clonal Selection Detection algorithm using nondominated sorting," 11th International IGTE Symposium
Algorithm for the V-BLAST System," Advances in Natural Computation, pp. on Numerical Field Calculation in Electrical Engineering , Seggauberg,
402-411, 2006. Austria, pp. 308-313, 2004.
[18] Carlos A. Coello Coello and Nareli Cruz Cortes, "An Approach to [38] F. G. Guimaraes, F. Campelo, R. R. Saldanha, H. Igarashi, R. H. C.
Solve Multiobjective Optimization Problems Based on an Artificial Immune Takahashi, and J. A. Ramirez, A multiobjective proposal for the TEAM
System," First International Conference on Artificial Immune Systems benchmark problem 22 IEEE Transactions on Magnetics, vol. 42, pp. 1471-
(ICARIS), University of Kent at Catenbury, UK, pp. 212-221, 2002. 1474, Apr, 2006.
[19] Charles Darwin. The Origin of Species by means of Natural [39] Fang Liu and Lan Luo, "Immune Clonal Selection Wavelet
Selection, or, The preservation of favored races in the struggle for life, Network Based Intrusion Detection," Proceedings Artificial Neural Networks:
Champaign : Project Gutenberg, 1859. Biological Inspirations - ICANN 2005, 15th International Conference,
Warsaw, Poland, pp. 331-336, 2005.
[20] Cheng Bo and Guo Zhenyu, "Adaptive Parallel Immune
Evolutionary Strategy," International Conference on Computational [40] Fang Liu, Qu Bo, and Chen Rongsheng, "Intrusion Detection
Intelligence and Security, pp. 304-307, 2006. Based on Immune Clonal Selection Algorithms," Australian Conference on
Artificial Intelligence - AI 2004: Advances in Artificial Intelligence, 2004.
[21] Cuiqin Hou, Licheng Jiao, Maoguo Gong, and Bin Lu, "Clone
Selection Based Multicast Routing Algorithm," Advances in Natural [41] Fang Liu, Qu Bo, and Chen Rongsheng, "A Novel Clonal
Computation, First International Conference, (ICNC 2005), Changsha, China, Selection Algorithm for Face Detection," AI 2006: Advances in Artificial
pp. 768-771, 2005. Intelligence, pp. 1226-1232, 2006.
[22] Daniel Stevens, Sanjoy Das, and Bala Natarajan, "A multi- [42] Fang Liu, Yuan Liu, Xi Chen, and Jin-shi Wang, "Multi-Agent
objective algorithm for DS-CDMA code design based on the clonal selection Immune Clonal Selection Algorithm Based Multicast Routing," Advances in
principle," Proceedings of the 2005 conference on Genetic and evolutionary Natural Computation, pp. 319-327, 2006.
computation, Washington DC, USA, pp. 2015-2020, 2005. [43] Frank Macfarlane Burnet, A modification of Jerne s theory of

[23] David B. Fogel. Evolutionary Computation : Toward a New antibody production using the concept of clonal selection Australian Journal
Philosophy of Machine Intelligence, New Jersey, USA : John Wiley & Sons, of Science, vol. 20, pp. 67-69, 1957.
Inc. 1995. [44] Frank Macfarlane Burnet. The clonal selection theory of acquired
[24] David E. Goldberg. Genetic Algorithms in Search, Optimization immunity, Nashville, Tennessee, U.S.A.: Vanderbilt University Press, 1959.
and Machine Learning, USA, Canada: Addison Wesley Publishing [45] Frank Macfarlane Burnet, Clonal selection and after Theoretical
Company, Inc., 1989. Immunology, vol. pp. 63-85, 1978.
[25] David E. Goldberg and J. Richardson, "Genetic algorithms with [46] Gaurav Marwah and Lois Boggess, "Artificial Immune Systems for
sharing for multimodal function optimization," Proceedings of the Second classification: Some issues," First International Conference on Artificial
International Conference on Genetic Algorithms on Genetic algorithms and Immune Systems, September 2002, pp. 149-153, 2002.
their application, Cambridge, MA, USA, pp. 41-49, 1987. [47] Hai-Feng Du , Li-Cheng Jiao, and Sun-An Wang, "Clonal operator
[26] Debraj De, Sonai Ray, and Amit Konar, "A Fuzzy Based Dynamic and antibody clone algorithms," Proceedings. 2002 International Conference
Routing Algorithm: A Q-Learning Intelligent Clonal Selection Approach with on Machine Learning and Cybernetics, pp. 506-510, 2002.
SPDE Pareto," ICCI 2004: International Conference on Computational [48] Haifeng Du, Licheng Jiao, Maoguo Gong, and Ruochen Liu,
Intelligence, Istanbul; Turkey, pp. 396-399, 2004. "Adaptive Dynamic Clone Selection Algorithms," Rough Sets and Current
[27] Donald E. Goodman and Lois C. Boggess, "The role of hypothesis Trends in Computing, pp. 768-773, 2004.
filtering in (AIRS), an artificial immune classifier," Intelligent Engineering [49] Haifeng Du, Xiaoyi Jin, Jian Zhuang, Licheng Jiao, and Sun'an
Systems Through Artificial Neural Networks, pp. 243-248, 2004. Wang, "Immune Clonal Selection Network," Advances in Artificial
[28] Donald Goodman, Lois Boggess, and Andrew Watkins, "Artificial Intelligence, 17th Australian Joint Conference on Artificial Intelligence,
Immune System Classification of Multiple-Class Problems," Intelligent Cairns, Australia, pp. 840-852, 2004.
Engineering Systems through Artificial Neural Networks: Smart Engineering [50] Hang Yu, Maoguo Gong, Licheng Jiao, and Bin Zhang, "Clonal
System Design: Neural Networks, pp. 179-184, 2002. Selection Algorithm with Immunologic Regulation for Function
[29] Donald Goodman, Lois Boggess, and Andrew Watkins, "An Optimization," Computational Intelligence and Security, International
Investigation into the Source of Power for AIRS, an Artificial Immune Conference, CIS 2005, Xi'an, China, pp. 858-863, 2005.
Classification System," Proceedings of the International Joint Conference on [51] Hong-yun Meng, Xiao-hua Zhang, and San-yang Liu, "A Novel
Neural Networks (IJCNN'03), pp. 1678-1683, 2003. Clonal Selection for Multi-modal Function Optimization," Advances in

CIS Technical Report 070209A February 2007 page 9 of 13

Natural Computation, pp. 63-72, 2006. [71] Jungwon Kim and Peter J. Bentley, "Towards an artificial immune
[52] Hongwei Dai, Yu Yang, Yanqiu Che, and Zheng Tang, "Clonal system for network intrusion detection: an investigation of clonal selection
Selection Theory Based Artificial Immune System and Its Application," with a negative selection operator," Proceedings of the 2001 Congress on
Neural Information Processing - Neurodynamic and Particle Swarm Evolutionary Computation, Seoul, South Korea, pp. 1244-1252, 2001.
Optimization, pp. 1071-1078, 2006. [72] Jungwon Kim and Peter J. Bentley, "Immune Memory in the
[53] Janna S. Hamaker and Lois Boggess, "Non-Euclidean distance Dynamic Clonal Selection Algorithm," Proceedings of the First International
measures in AIRS, an artificial immune classification system," Proceedings of Conference on Artificial Immune Systems (ICARIS), University of Kent,
the 2004 Congress of Evolutionary Computation (CEC2004), pp. 1067-1073, Canterbury, UK, pp. 57-65, 2002.
2004. [73] Jungwon Kim and Peter J. Bentley, "A Model of Gene Library
[54] Jennifer White and Simon M. Garrett, "Improved Pattern Evolution in the Dynamic Clonal Selection Algorithm," Proceedings of the
Recognition with Artificial Clonal Selection?," Proceedings Artificial Immune First International Conference on Artificial Immune Systems (ICARIS),
Systems: Second International Conference, ICARIS 2003, Edinburgh, UK, pp. University of Kent at Canterbury, UK, pp. 175-182, 2002.
181-193, 2003. [74] Jungwon Kim and Peter J. Bentley, "Towards an Artificial
[55] Ji-Qing Xian, Feng-Hua Lang, and Xian-Lun Tang, "A novel Immune System for Network Intrusion Detection: An Investigation of
intrusion detection method based on clonal selection clustering algorithm," Dynamic Clonal Selection," Congress on Evolutionary Computation (CEC-
Proceedings of 2005 International Conference on Machine Learning and 2002), Honolulu, USA, pp. 1015-1020, 2002.
Cybernetics, pp. 3905-3910, 2005. [75] K. Deb and David E. Goldberg, "An Investigation of Niche and
[56] Jiang-qiang Hu, Chen Guo, Tie-shan Li, and Jian-chuan Yin, Species Formation in Genetic Function Optimization," Proceedings of the
"Adaptive Clonal Selection with Elitism-Guided Crossover for Function Third International Conference on Genetic Algorithms, George Mason
Optimization," ICICIC '06. First International Conference on Innovative University, Fairfax, Virginia, USA, pp. 42-50, 1989.
Computing, Information and Control, pp. 206-209 , 2006. [76] Kazuyuki Mori, Makoto Tsukiyama, and Toyoo Fukuda, Immune
[57] Jing Li, Licheng Jiao, Maoguo Gong, and Wuhong He, "Lamarkian algorithm with searching diversity and its application to resource allocation
Clonal Selection Algorithm for CDMA Multiuser Detection over Multi-Path problem The Transactions of the Institute of Electrical Engineers of Japan,
Channels," International Conference on Neural Networks and Brain, 2005. vol. 113-C, pp. 872-878, 1993.
ICNN&B '05, pp. 601-606, 2005. [77] Kemal Polat , Sadik Kara, Fatma Latifoglu, and Salih G nes, "A 

[58] Jing Wang, Xiao-hua Zhang, and Licheng Jiao, "Integrated the Novel Approach to Resource Allocation Mechanism in Artificial Immune
Simplified Interpolation and Clonal Selection into the Particle Swarm Recognition System: Fuzzy Resource Allocation Mechanism and Application
Optimization for Optimization Problems," Simulated Evolution and Learning, to Diagnosis of Atherosclerosis," Artificial Immune Systems, pp. 244-255,
pp. 433-440, 2006. 2006.
[59] Joanne H. Walker and Simon M. Garrett, "Dynamic Function [78] Kemal Polat , Seral Sahan, Halife Kodaz, and Salih G nes, "A


Optimisation: Comparing the Performance of Clonal Selection and New Classification Method for Breast Cancer Diagnosis: Feature Selection
Evolutionary Strategies," ICARIS-2003, Edinburgh, pp. 273-284, 2003. Artificial Immune Recognition System (FS-AIRS)," Advances in Natural
Computation, pp. 830-838, 2005.
[60] Johnny Kelsey, J. Timmis, and A. Hone, "Chasing chaos," The
2003 Congress on Evolutionary Computation, (CEC '03), pp. 413-419, 2003. [79] Kemal Polat , Seral Sahan, Halife Kodaz, and Salih G nes, 

"Outdoor Image Classification Using Artificial Immune Recognition System

[61] Johnny Kelsey and Jon Timmis, "Immune Inspired Somatic (AIRS) with Performance Evaluation by Fuzzy Resource Allocation
Contiguous Hypermutation for Function Optimisation," Proceedings, Part I
Mechanism," Computer Analysis of Images and Patterns, pp. 81-87,
Genetic and Evolutionary Computation Conference (GECCO 2003), Chicago, 2005.
IL, USA, pp. 207-218, 2003.
[80] Kemal Polat , Seral Sahan, and Salih G nes, A new method to
[62] Jon Timmis, C. Edmonds, and Johnny Kelsey, "Assessing the


medical diagnosis: Artificial immune recognition system (AIRS) with fuzzy

Performance of Two Immune Inspired Algorithms and a Hybrid Genetic
weighted pre-processing and application to ECG arrhythmia Expert Systems
Algorithm for Function Optimisation," Proceedings of the Congress on
with Applications , vol. 31, pp. 264-269, Aug, 2006.
Evolutionary Computation (CEC04), Potland, Oregon. USA, pp. 1044-1051,
2004. [81] Kerim Guney, B. Babayigit, and A. Akdagli, "Position only pattern
nulling of linear antenna array by using a clonal selection algorithm
[63] Jorge Lu s Machado do Amaral, Jos Franco Machado do Amaral,
(CLONALG)," Electrical Engineering (Archiv fur Elektrotechnik), 2007.

Ricardo Tanscheit, Marco Aur lio Cavalcanti Pacheco, and Antonio Carneiro
[82] L. Xu, M. Y. Chow, and J. Timmis. L. S. Taylor, Power

de Mesquita Filho, "Tuning evolvable PID controllers through a clonal

selection algorithm," Proceedings of the 2005 NASA/DoD Conference on Distribution Outage Cause Identification With Imbalanced Data Using
Evolvable Hardware , pp. 30-33, 2005. Artificial Immune Recognition System (AIRS) Algorithm IEEE Transactions
on Power Systems, vol. 22, pp. 198-204, Feb, 2007.
[64] Juan Carlos Galeano, Ang lica Veloza-Suan, and Fabio A.
[83] Le Xu, Mo-Yuen Chow, J. Timmis, L. S. Taylor, and A. Watkins,

Gonz lez, "A comparative analysis of artificial immune network models,"

"On the Investigation of Artificial Immune Systems on Imbalanced Data


Proceedings of the 2005 conference on Genetic and evolutionary computation

(GECCO'05), Washington DC, USA, pp. 361-368, 2005. Classification for Power Distribution System Fault Cause Identification,"
CEC 2006. IEEE Congress on Evolutionary Computation, pp. 522-527 ,
[65] Julie Greensmith, New Frontiers For An Artificial Immune
2006.
System 2003. University of Leeds.
[84] Leandro N. de Castro and Fernando J. Von Zuben, "The Clonal
[66] Julie Greensmith and Steve Cayzer, "An Artificial Immune System Selection Algorithm with Engineering Applications," Proceedings of the
Approach to Semantic Document Classification," Proceedings Artificial
Genetic and Evolutionary Computation Conference (GECCO '00), Workshop
Immune Systems: Second International Conference, ICARIS 2003, Edinburgh, on Artificial Immune Systems and Their Applications, Las Vegas, Nevada,
UK, pp. 136-146, 2003.
USA, pp. 36-37, 2000.
[67] Jungan Chen, Dongyong Yang, and Feng Liang, "Static Clonal
[85] Leandro N. de Castro and Fernando J. Von Zuben, Learning and
Selection Algorithm Based on Match Range Model," Advances in Applied optimization using the clonal selection principle IEEE Transactions on
Artificial Intelligence, pp. 859-868, 2006.
Evolutionary Computation, vol. 6, pp. 239-251, Jun, 2002.
[68] Jungwon Kim and Peter Bentley, Immune Memory and Gene
[86] Leandro N. de Castro and Fernando Jos Von Zuben , "Artificial
Library Evolution in the Dynamic Clonal Selection Algorithm Genetic

Immune Systems - Part I: Basic Theory and Applications," Department of

Programming and Evolvable Machines, vol. 5, pp. 361-391, Dec, 2004.
Computer Engineering and Industrial Automation, School of Electrical and
[69] Jungwon Kim and Peter J. Bentley, "Negative Selection and Computer Engineering, State University of Campinas, Brazil, TR DCA 01/99,
Niching by an Artificial Immune System for Network Intrusion Detection," A Dec 1999.
late-breaking paper, Genetic and Evolutionary Computation Conference
[87] Leandro N. de Castro and Jon Timmis, "An artificial immune
(GECCO '99), Orlando, Florida, USA, 1999.
network for multimodal function optimization," Proceedings of the 2002
[70] Jungwon Kim and Peter J. Bentley, "An Evaluation of Negative Congress on Evolutionary Computation (CEC '02), Honolulu, HI, USA, pp.
Selection in an Artificial Immune System for Network Intrusion Detection," 699-704, 2002.
Proceedings of the Genetic and Evolutionary Computation Conference
[88] Leandro N. de Castro and Jon Timmis. Artificial Immune Systems:
(GECCO-2001), pp. 1330-1337, 2001. A new computational intelligence approach, Great Britain: Springer-Verlag,

CIS Technical Report 070209A February 2007 page 10 of 13

2002. Shih, "Power Transformer Fault Diagnosis Using Support Vector Machines
[89] Li Jie, Gao Xinbo, and Jiao Li-Cheng, "A CSA-based clustering and Artificial Neural Networks with Clonal Selection Algorithms
algorithm for large data sets with mixed numeric and categorical values," Optimization," Knowledge-Based Intelligent Information and Engineering
Fifth World Congress on Intelligent Control and Automation, WCICA 2004. Systems, pp. 179-186, 2006.
pp. 2303-2307, 2004. [108] Ming-Yuan Cho, Tsair-Fwu Lee, Shih-Wei Kau, Chin-Shiuh
[90] Licheng Jiao, Maoguo Gong, Ronghua Shang, Haifeng Du, and Shieh, and Chao-Ji Chou, "Fault Diagnosis of Power Transformers Using
Bin Lu, "Clonal Selection with Immune Dominance and Anergy Based SVM/ANN with Clonal Selection Algorithm for Features and Kernel
Multiobjective Optimization," Proceedings of the Third International Parameters Selection," ICICIC '06. First International Conference on
Conference Evolutionary Multi-Criterion Optimization (EMO 2005), Innovative Computing, Information and Control, 2006. pp. 26-30, 2006.
Guanajuato, M xico, pp. 474-489, 2005.

[109] Nareli Cruz Cort s and Carlos A. Coello Coello, "Multiobjective

[91] Lin Hao, Maoguo Gong, Yifei Sun, and Jin Pan, "Niching Clonal Optimization Using Ideas from the Clonal Selection Principle," Proceedings,
Selection Algorithm for Multimodal Function Optimization," Advances in Part I Genetic and Evolutionary Computation Conference (GECCO 2003),
Natural Computation, Second International Conference, ICNC 2006, Xi'an, Chicago, IL, USA, pp. 158-170, 2003.
China, pp. 820-827, 2006. [110] Nareli Cruz-Cort s, Daniel Trejo-P rez, and Carlos A. Coello

[92] Lingjun Meng, Peter van der Putten, and Haiyang Wan, "A Coello, "Handling Constraints in Global Optimization using an Artificial
Comprehensive Benchmark of the Artificial Immune Recognition System Immune System," Proceedings of the 4th International Conference on
(AIRS)," Advanced Data Mining and Applications, pp. 575-582, 2005. Artificial Immune Systems (ICARIS 2005) , Banff, Alberta, Canada , pp. 234-
247, 2005.
[93] Liu Fang and Li Chaoyang, "A multiview face recognition based
on combined feature with clonal selection," Proceedings. ICSP '04. 2004 7th [111] P. Bull, A. Knowles, G. Tedesco, and A. Hone, "Diophantine
International Conference on Signal Processing, pp. 1380-1384, 2004. Benchmarks for the B-Cell Algorithm," Proceedings 5th International
Conference on Artificial Immune Systems (ICARIS 2006), Instituto
[94] Liu Fang and Zhao Jing: "An Immune Clonal Selection Scheduling
Gulbenkian de Ci ncia, Oeiras, Portugal, 2006.
Algorithm for Input-Queued Switches," Simulated Evolution and Learning,
pp. 790-797, 2006. [112] Philip Leder. The Genetics of Antibody Diversity . In:
Immunology Recognition and Response - Readings from scientific american
[95] Liu Haibo, Gu Guochang, Shen Jing, and Fu Yan, "A fast clonal magazine, ed. William E. Paul. New York: W. H. Freeman and Company,
selection algorithm for constructing an immune neural network,"
1987.pp. 20-34.
PROCEEDINGS OF THE 11TH JOINT INTERNATIONAL COMPUTER
CONFERENCE - JICC 2005, Chongqing, China, 2005. [113] Qiaoling Wang, Changhong Wang, and X. Z. Gao, "A Hybrid
Optimization Algorithm based on Clonal Selection Principle and Particle
[96] Liu Ruochen, Du Haifeng, and Jiao Licheng, "Immunity clonal
Swarm Intelligence," Sixth International Conference on Intelligent Systems
strategies," Proceedings. Fifth International Conference on Computational Design and Applications, ISDA '06. pp. 975-979, 2006.
Intelligence and Multimedia Applications, ICCIMA 2003. pp. 290-295, 2003.
[114] Richard G. Weinand, Somatic mutation, affinity maturation and
[97] Lois Boggess and Janna S. Hamaker, "The Effect of Irrelevant
the antibody repertoire: a computer model Journal of Theoretical Biology,
Features on AIRS, an Artificial Immune-Based Classifier," Intelligent vol. 143, pp. 343-382, 1990.
Engineering Systems through Artificial Neural Networks (ANNIE), pp. 219-
224, 2003. [115] Ron Hightower, Stephanie Forrest, and Alan S. Perelson, "The
Baldwin Effect in the Immune System: Learning by Somatic Hypermutation,"
[98] Ma Jie, Gao Hong-yuan, and Diao Ming, "Multiuser Detection Adaptive Individuals in Evolving Populations: Models and Algorithms, pp.
Using the Clonal Selection Algorithm and Hopfield Neural Network,"
159-167, 1995.
International Conference on Communications, Circuits and Systems
Proceedings, pp. 739-743 , 2006. [116] Ronghua Shang, Licheng Jiao, Maoguo Gong, and Bin Lu, "Clonal
Selection Algorithm for Dynamic Multiobjective Optimization,"
[99] Ma Li, Bai Lin, Jiao Li-cheng, and Chen Chang-guo, "Intrusion
Computational Intelligence and Security, International Conference, CIS 2005,
Detection Based on Adaptive Polyclonal Clustering," International
Xi'an, China, pp. 846-851, 2005.
Conference on Computational Intelligence and Security, pp. 598-603, 2006.
[117] Ruo-Chen Liu, Li-Cheng Jiao, and Hai-Feng Du, Clonal Strategy
[100] Maoguo Gong, Haifeng Du1, Licheng Jiao, and Ling Wang,
Algorithm based on the Immune Memory Journal of Computer Science and
"Immune Clonal Selection Algorithm for Multiuser Detection in DS-CDMA
Technology, vol. 20, pp. 728-734, 2005.
Systems," AI 2004: Advances in Artificial Intelligence, 17th Australian Joint
Conference on Artificial Intelligence, Cairns, Australia, pp. 1219-1225, 2004. [118] Ruochen Liu, Li Chen, and Shuang Wang, "Immune Clonal
Strategies Based on Three Mutation Methods," Advances in Natural
[101] Maoguo Gong, Licheng Jiao, Haifeng Du, Bin Lu, and Wentao
Computation, pp. 114-121, 2006.
Huang, "IFCPA: Immune Forgetting Clonal Programming Algorithm for
Large Parameter Optimization Problems," Advances in Natural Computation, [119] Ruochen Liu, Licheng Jiao, and Haifeng Du, "Adaptive immune
First International Conference, ICNC 2005, Changsha, China, pp. 826-829, clonal strategy algorithm," Proceedings. ICSP '04. 2004 7th International
2005. Conference on Signal Processing, pp. 1554-1557, 2004.
[102] Maoguo Gong, Licheng Jiao, Haifeng Du, Ronghua Shang, and [120] Samir W. Mahfoud, "A Comparison of Parallel and Sequential
Bin Lu, "Performance assessment of an artificial immune system Niching Methods," Proceedings of the Sixth International Conference on
multiobjective optimizer by two improved metrics," Proceedings of the 2005 Genetic Algorithms, pp. 136-143 , 1995.
conference on Genetic and evolutionary computation, Washington DC, USA, [121] Shuiping Gou, Licheng Jiao, Yangyang Li, and Qing Li, "Kernel
pp. 373-374, 2005. Matching Pursuit Based on Immune Clonal Algorithm for Image
[103] Maoguo Gong, Ling Wang, Licheng Jiao, and Haifeng Du, "An Recognition," Simulated Evolution and Learning, pp. 26-33, 2006.
artificial immune system algorithm for CDMA multiuser detection over multi- [122] Simon M. Garrett, "Parameter-free, adaptive clonal selection,"
path channels," Proceedings of the 2005 conference on Genetic and Congress on Evolutionary Computing (CEC 2004), Portland Oregon, USA,
evolutionary computation, Washington DC, USA, pp. 2105-2111, 2005. pp. 1052-1058, 2004.
[104] Mario Villalobos-Arias, Carlos A. Coello Coello, and On simo
[123] Stephanie Forrest, Brenda Javornik, Robert E. Smith, and Alan S.
Hern ndez-Lerma, "Convergence Analysis of a Multiobjective Artificial
 Perelson, Using Genetic Algorithms to Explore Pattern Recognition in the
Immune System Algorithm," Proceedings Artificial Immune Systems: Third Immune System Evolutionary Computation, vol. 1, pp. 191-211, 1993.
International Conference, ICARIS 2004, Catania, Sicily, Italy, pp. 226-235, [124] Thomas Back, David B Fogel, and Zbigniew Michalwicz.
2004. Evolutionary Computation 1 - Basic Algorithms and Operators, Bristol, UK:
[105] Meiyi Li, Zixing Cai, Yuexiang Shi, and Pingan Gao, "A Hybrid Institute of Physics (IoP) Publishing, 2000.
Immune Evolutionary Computation Based on Immunity and Clonal Selection [125] Thomas Back, David B Fogel, and Zbigniew Michalwicz.
for Concurrent Mapping and Localization," Advances in Natural Evolutionary Computation 2 - Advanced Algorithms and Operators, Bristol,
Computation, pp. 1308-1311, 2005. UK: Institute of Physics (IoP) Publishing, 2000.
[106] Melanie Mitchell and John H. Holland, "When Will a Genetic [126] Toyoo Fukuda, Kazuyuki Mori, and Makoto Tsukiyama, Immune
Algorithm Outperform Hill Climbing?," Proceedings of the 5th International networks using genetic algorithm for adaptive production scheduling 15th
Conference on Genetic Algorithms, Urbana-Champaign, IL, USA, pp. 647 , IFAC World Congress, vol. 3, pp. 57-60, 1993.
1993.
[127] Tsair-Fwu Lee, Ming-Yuan Cho, Chin-Shiuh Shieh, Hong-Jen Lee,
[107] Ming-Yuan Cho, Tsair-Fwu Lee, Shih-Wei Gau, and Ching-Nan

CIS Technical Report 070209A February 2007 page 11 of 13

and Fu-Min Fang, "Diagnosis of Incipient Fault of Power Transformers Using [145] Vincenzo Cutello and Giuseppe Nicosia, "A Clonal Selection
SVM with Clonal Selection Algorithms Optimization," Foundations of Algorithm for Coloring, Hitting Set and Satisfiability Problems," Proceedings
Intelligent Systems, pp. 580-590, 2006. of the Internation Workshop on Natural and Artificial Immune Systems, NAIS
[128] U. Garain, M. P. Chakraborty, and D. D. Majumder, "Improvement 2005, Italy, pp. 324-337, 2006.
of OCR Accuracy by Similar Character Pair Discrimination: an Approach [146] Vincenzo Cutello, Giuseppe Nicosia, and Mario Pavone, "A
based on Artificial Immune System," 18th International Conference on Hybrid Immune Algorithm with Information Gain for the Graph Coloring
Pattern Recognition, (ICPR 2006) , pp. 1046-1049, 2006. Problem," Proceedings, Part I: Genetic and Evolutionary Computation
[129] Utpal Garain, Mangal P. Chakraborty, and Dipankar Dasgupta, Conference (GECCO 2003), Chicago, IL, USA, pp. 171-182, 2003.
"Recognition of Handwritten Indic Script Using Clonal Selection Algorithm," [147] Vincenzo Cutello, Giuseppe Nicosia, and Mario Pavone,
Proceedings Artificial Immune Systems, 5th International Conference "Exploring the Capability of Immune Algorithms: A Characterization of
(ICARIS 2006), Oeiras, Portugal, pp. 256-266, 2006. Hypermutation Operators," Proceedings Artificial Immune Systems: Third
[130] V. Cutello , D. Lee, S. Leone, G. Nicosia, and M. Pavone, "Clonal International Conference, ICARIS 2004, Catania, Sicily, Italy, pp. 263-276,
Selection Algorithm with Dynamic Population Size for Bimodal Search 2004.
Spaces," 2nd International Conference on Natural Computation (ICNC), [148] Vincenzo Cutello, Giuseppe Nicosia, and Mario Pavone, "An
Xi'an, China, pp. 949-958, 2006. immune algorithm with hyper-macromutations for the Dill's 2D hydrophobic-
[131] V. Cutello, G. Narzisi, G. Nicosia, and M. Pavone, "An hydrophilic model," Congress on Evolutionary Computation, (CEC2004), pp.
Immunological Algorithm for Global Numerical Optimization," 7th 1074-1080, 2004.
International Conference on Artificial Evolution (EA), Lille, France, pp. 284- [149] Weisheng Dong, Guangming Shi, and Li Zhang, Immune memory
295, 2005. clonal selection algorithms for designing stack filters Connectionists:
[132] V. Cutello, G. Nicosia, and E. Pavia, "A Parallel Immune Neurocomputing, vol. 70, pp. 777-784, 2007.
Algorithm for Global Optimization," The 7th Int. Conf. on Intelligent [150] Wenping Ma, Licheng Jiao, Maoguo Gong, and Fang Liu, "An
Information Systems (IIS 2006), Ustron, Poland, pp. 467-475, 2006. Novel Artificial Immune Systems Multi-objective Optimization Algorithm for
[133] V. Cutello, G. Nicosia, M. Pavone, and G. Narzisi, "Real coded 0/1 Knapsack Problems," Computational Intelligence and Security,
clonal selection algorithm for unconstrained global optimization using a International Conference, CIS 2005, Xi'an, China, pp. 793-798, 2005.
hybrid inversely proportional hypermutation operator," Proceedings of the [151] Wenping Ma, Licheng Jiao, Maoguo Gong, and Ronghua Shang,
2006 ACM symposium on Applied computing, Dijon, France, pp. 950-954, "Immune Clonal Selection Evolutionary Strategy for Constrained
2006. Optimization," PRICAI 2006: Trends in Artificial Intelligence, 9th Pacific
[134] V. Cutello , G. Nicosia, M. Pavone, and J. Timmis, An Immune Rim International Conference on Artificial Intelligence, Guilin, China, pp.
Algorithm for Protein Structure Prediction on Lattice Models IEEE 661-670.
Transactions on Evolutionary Computation, vol. 2006. [152] Wenping Ma and Ronghua Shang, "SAR Image Classification
[135] V.I. Litvinenko, P. I Bidyuk, J. N. Bardachov, V. G. Sherstjuk, and Based on Clonal Selection Algorithm," Advances in Natural Computation, pp.
A. A. Fefelov, "Combining Clonal Selection Algorithm and Gene Expression 927-934, 2006.
Programming for Time Series Prediction," Proceedings of The Third [153] William E. Paul. Immunology - Recognition and Response :
Workshop 2005 IEEE Intelligent Data Acquisition and Advanced Computing Readings from Scientific American magazine / edited by William E. Paul.,
Systems: Technology and Applications, pp. 133-138 , 2005. USA: Scientific American Inc., W. H. Freeman and Company, 1991.
[136] V. K. Karakasis and A. Stafylopatis, "Data Mining based on Gene [154] Wuhong He, Haifeng Du, Licheng Jiao, and Jing Li, "Lamarckian
Expression Programming and Clonal Selection," IEEE Congress on Clonal Selection Algorithm Based Function Optimization," 8th International
Evolutionary Computation, CEC 2006. pp. 514-521, 2006. Work-Conference on Computational Intelligence And Bioinspired Systems,
[137] Vincenzo Cutello, D. Lee, Giuseppe Nicosia, Mario Pavone, and I. pp. 91-98, 2005.
Prizzi, "Aligning Multiple Protein Sequences by Hybrid Clonal Selection [155] Wuhong He, Haifeng Du, Licheng Jiao, and Jing Li, "Lamarckian
Algorithm with Insert-Remove-Gaps and BlockShuffling Operators," Clonal Selection Algorithm with Application," Artificial Neural Networks:
Artificial Immune Systems, 5th International Conference, ICARIS 2006, Biological Inspirations ICANN 2005 (Evolutionary and Other Biological
Oeiras, Portugal, pp. 321-334, 2006. Inspirations), pp. 317-322, 2005.
[138] Vincenzo Cutello, G. Morelli, Giuseppe Nicosia, and Mario [156] X.L. Wang and M. Mahfouf, "ACSAMO: An Adaptive
Pavone, "Immune Algorithms with Aging Operators for the String Folding Multiobjective Optimization Algorithm using the Clonal Selection Principle
Problem and the Protein Folding Problem ," Proceedings Evolutionary ," European Symposium on Nature-Inspired Smart Information Systems
Computation in Combinatorial Optimization: 5th European Conference (NiSIS 2005), Albufeira, Portugal, 2005.
(EvoCOP 2005), Lausanne, Switzerland, pp. 80-90, 2005. [157] X. Wang, "Clonal selection algorithm in power filter
[139] Vincenzo Cutello, G. Narzisi, Giuseppe Nicosia, Mario Pavone, optimization," Proceedings of the 2005 IEEE Mid-Summer Workshop on Soft
and G. Sorace, "How to Escape Traps using Clonal Selection Algorithms," Computing in Industrial Applications, SMCia/05. pp. 122-127, 2005.
The First International Conference on Informatics in Control, Automation [158] X. Wang, X. Z. Gao, and S. J. Ovaska, "A hybrid optimization
and Robotics, ICINCO 2004, Setubal, Portugal, pp. 322-326, 2004. algorithm in power filter design," 32nd Annual Conference of IEEE Industrial
[140] Vincenzo Cutello, Giuseppe Narzisi, and Giuseppe Nicosia, A Electronics Society, IECON 2005. 2005.
multi-objective evolutionary approach to the protein structure prediction [159] Xiangrong Zhang and Licheng Jiao, "Simultaneous Feature
problem Journal of the Royal Society Interface, vol. 3, pp. 139-151, 2006. Selection and Parameters Optimization for SVM by Immune Clonal
[141] Vincenzo Cutello, Giuseppe Narzisi, Giuseppe Nicosia, and Mario Algorithm," Advances in Natural Computation, pp. 905-912, 2005.
Pavone, "Clonal Selection Algorithms: A Comparative Case Study using [160] Xiangrong Zhang, Shuang Wang, Tan Shan, and Licheng Jiao,
Effective Mutation Potentials," Proceedings Artificial Immune Systems: 4th "Radar target recognition using SVMs with a wrapper feature selection driven
International Conference, ICARIS 2005, Banff, Alberta, Canada, pp. 13-28, by immune clonal algorithm," ESANN 2005, 13th European Symposium on
2005. Artificial Neural Networks, Bruges, Belgium, pp. 577-582, 2005.
[142] Vincenzo Cutello and Giuseppe Nicosia, "An Immunological [161] Xiangrong Zhang, Shuang Wang, Tan Shan, and Licheng Jiao,
Approach to Combinatorial Optimization Problems," Proceedings of the 8th "Selective SVMs Ensemble Driven by Immune Clonal Algorithm,"
Ibero-American Conference on AI: Advances in Artificial Intelligence, Applications on Evolutionary Computing, pp. 325-333, 2005.
Seville, Spain, pp. 361-370, 2002.
[162] Xiangrong Zhang, Tan Shan, and Licheng Jiao, "SAR Image
[143] Vincenzo Cutello and Giuseppe Nicosia, "Multiple learning using Classification Based on Immune Clonal Feature Selection," Image Analysis
immune algorithms," Proceedings of 4th International Conference on Recent and Recognition: International Conference, ICIAR 2004, Porto, Portugal,
Advances in Soft Computing, RASC 2002, Nottingham, UK, pp. 102-107, 2004.
2002.
[163] Xiao-hui Yang, Licheng Jiao, Yutao Qi, and Hai-yan Jin,
[144] Vincenzo Cutello and Giuseppe Nicosia. Chapter VI. The Clonal "Multifocus Image Fusion Based on Multiwavelet and Immune Clonal
Selection Principle for In Silico and In Vivo Computing. In: Recent Selection," Advances in Natural Computation, pp. 805-815, 2006.
Developments in Biologically Inspired Computing, eds. Leandro Nunes de
[164] Xiaomeng Bian and Jiaju Qiu, "Adaptive Clonal Algorithm and Its
Castro and Fernando J. Von Zuben. Hershey, London, Melbourne, Singapore:
Idea Group Publishing, 2005.pp. 104-146. Application for Optimal PMU Placement," Proceedings of the International
Conference on Communications, Circuits and Systems, pp. 2102-2106 , 2006.

CIS Technical Report 070209A February 2007 page 12 of 13

 [165] Xiaoying Pan, Fang Liu, and Licheng Jiao, "A Dynamic Clonal [174] Yanping Lv, Shaozi Li, Shuili Chen, Qingshan Jiang, and
Selection Algorithm for Project Optimization Scheduling," Simulated Wenzhong Guo, "Particle Swarm Optimization Based on Information
Evolution and Learning, pp. 821-828, 2006. Diffusion and Clonal Selection," Simulated Evolution and Learning, pp. 521-
[166] Xin Jin, Rongfang Bie, and X. Z. Gao, "An Artificial Immune 528, 2006.
Recognition System-based Approach to Software Engineering Management [175] Yi-Hui Su, Wen-Jye Shyr, and Te-Jen Su, "Optimal Design Using
with Software Metrics Selection," Proceedings of the Sixth International Clonal Selection Algorithm," Knowledge-Based Intelligent Information and
Conference on Intelligent Systems Design and Applications (ISDA'06) , pp. Engineering Systems, pp. 604-610, 2004.
523-528 , 2006. [176] Yifei Sun, Maoguo Gong, Lin Hao, and Licheng Jiao, "Clonal
[167] Xinbo Gao, Juxia Gu, and Jie Li, De-interlacing algorithms based Selection Algorithm with Search Space Expansion Scheme for Global
on motion compensation IEEE Transactions on Consumer Electronics, vol. Function Optimization," Advances in Natural Computation, Second
51, pp. 589-599, May, 2005. International Conference, ICNC 2006, Xi'an, China, pp. 838-847, 2006.
[168] Xing Quan Zuo and Y. S. Fan, " The chaos artificial immune [177] Ying Yu and Chao-Zhen Hou, "A clonal selection algorithm by
algorithm and its application to RBF neuro-fuzzy controller design," IEEE using learning operator," Proceedings of 2004 International Conference on
International Conference on Systems, Man and Cybernetics, pp. 2809-2814, Machine Learning and Cybernetics, pp. 2924-2929, 2004.
2003. [178] Yutao Qi, Xiaoying Pan, Fang Liu, and Licheng Jiao, "A Strategy
[169] Yajing Zhang and Chaozhen Hou, "A clone selection algorithm of Mutation History Learning in Immune Clonal Selection Algorithm,"
with niching strategy inspiring by biological immune principles for change Simulated Evolution and Learning, pp. 72-79, 2006.
detection," IEEE International Symposium on Intelligent Control, pp. 1000- [179] Z.X. Ong, J.C. Tay, and C.K. Kwoh, "Applying the Clonal
1005, 2003. Selection Principle to Find Flexible Job-Shop Schedules," Proceedings of the
[170] Yanfei Zhong, Liangpei Zhang, Bo Huang, and Pingxiang Li, An 4th International Conference on Artificial Immune Systems, pp. 442-455,
unsupervised artificial immune classifier for multi/hyperspectral remote 2005.
sensing imagery IEEE Transactions on Geoscience and Remote Sensing, vol. [180] Zhuo-Yue Song, X.Z. Gao, Xian-Lin Huang, and H. S. Lin, "A
44, pp. 420-431, Feb, 2006. Modified Immune Optimization Algorithm," International Conference on
[171] Yanfei Zhong, Liangpei Zhang, and Pingxiang Li, "Multispectral Machine Learning and Cybernetics, pp. 2184-2189, 2006.
remote sensing image classification based on simulated annealing clonal [181] Zongxin Jin, Guangyuan Liu, and Wanhui Wen, "Clonal Selection
selection algorithm," Proceedings. 2005 IEEE International Geoscience and Algorithm with Hyper Mutation and Spatial Clone Extension," International
Remote Sensing Symposium, IGARSS '05. pp. 3745-3748, 2005. Conference on Computational Intelligence and Security, pp. 402-405, 2006.
[172] Yangyang Li and Licheng Jiao, "Quantum-Inspired Immune
Clonal Algorithm," Artificial Immune Systems, pp. 304-317, 2005.
[173] Yangyang Li, Licheng Jiao, and Shuiping Gou, "Quantum-Inspired
Immune Clonal Algorithm for Multiuser Detection in DS-CDMA Systems,"
Simulated Evolution and Learning, pp. 80-87, 2006.

CIS Technical Report 070209A February 2007 page 13 of 13