Artificial Immune Systems

Fugue is the result of a collaboration between artist, musician and computer scientists. The result is an on-going project which provides a new way of communicating complex scientific ideas to any audience. Immersive virtual reality and sound provide an interactive audiovisual interface to the dynamics of a complex system – for this work, an artificial immune system. Participants are able to see and interact with immune cells flowing through a lymphatic vessel and understand how the complex dynamics of the whole are produced by local interactions of viruses, B cells, antibodies, dendritic cells and clotting platelets.

This paper presents a comparative study of two important Clonal Selection Algorithms (CSAs): CLONALG and opt-IA. To deeply understand the performance of both algorithms, we deal with four different classes of problems: toy problems (one-counting and trap functions), pattern recognition, numerical optimization problems and NP-complete problem (the 2D HP model for protein structure prediction problem). Two possible versions of CLONALG have been implemented and tested. The experimental results show a global better performance of opt-IA with respect to CLONALG. Considering the results obtained, we can claim that CSAs represent a new class of Evolutionary Algorithms for effectively performing searching, learning and optimization tasks.

The purpose of this paper is three-fold. Firstly, it aims to demonstrate empirically that networks evolved using different shaped recognition regions in a real-valued shape-space exhibit different dynamics during their formation, and vary in both their capabilities to tolerate antigens and in their memory capacity. Secondly, the paper serves as a useful comparison to previous published work which investigated the properties of a network evolving in a simple, small Hamming shape-space. This work represents the first steps in a proper analysis of a real-valued shape-space with differing recognition shapes. Finally, and perhaps most importantly, the experiments presented illustrate the importance of paying careful attention to the choice of recognition region and algorithm parameters when applying an AIS based on a network-model to practical problems.

Many bio-inspired algorithms (evolutionary algorithms, artificial immune systems, particle swarm optimisation, ant colony optimisation,...) are based on populations of agents. Stepney

et al

[2005] argue for the use of conceptual frameworks and meta-frameworks to capture the principles and commonalities underlying these, and other bio-inspired algorithms. Here we outline a generic framework that captures a collection of population-based algorithms, allowing commonalities to be factored out, and properties previously thought particular to one class of algorithms to be applied uniformly across all the algorithms. We then describe a prototype proof-of-concept implementation of this framework on a small grid of FPGA (field programmable gate array) chips, thus demonstrating a generic architecture for both parallelism (on a single chip) and distribution (across the grid of chips) of the algorithms.

An Artificial Immune System-based Dynamic Routing (AISDR) framework is engineered through the adoption of the characteristics that are analogous to human immune system for solving dynamic routing problems. The framework covers the profound features on recognition, selection, learning, memory, and adaptation capabilities. An AISDR algorithm is developed that incorporates the features of clonal selection, affinity maturation, and immunological memory features. Simulation study is carried out to evaluate the performance of the algorithm in the global shipment operation.

We propose to develop a theoretical basis and experimental simulator of the first Immune-Computer (IC) as a new kind of biomolecular computer. This IC will be able to control a fragment of the natural immune system in an autonomous and intelligent manner. Such control has proved unobtainable with other methods.

Artificial Immune Systems (AIS) have been shown to be useful, practical and realisable approaches to real-world problems. Most AIS implementations are based around a canonical algorithm such as clonotypic learning, which we may think of as individual, lifetime learning. Yet a species also learns.

Gene libraries

are often thought of as a biological mechanism for generating combinatorial diversity of antibodies. However, they also bias the antibody creation process, so that they can be viewed as a way of guiding the lifetime learning mechanisms. Over time, the gene libraries in a species will evolve to an appropriate bias for the expected environment (based on species memory). Thus gene libraries are a form of

meta-learning

which could be useful for AIS. Yet they are hardly ever used. In this paper we consider some of the possible benefits and implications of incorporating the evolution of gene libraries into AIS practice. We examine some of the issues that must be considered if the implementation is to be successful and beneficial.

The response to the title would simply be that the state of the organism has changed between the first and the seventh glass and that, before the seventh, this state was much closer to some kind of “homeostatic limit”. Although the external impact i.e. the glass of wine is identical in both cases, the reaction of the receptive organism might be different, depending on its current state: accept the first glass then reject the seventh. It is the couple “wine and current state of the organism” which is important here and not just the wine. Introducing this paper, I will attempt to clarify the famous self-nonself controversy by referring attentively to the debate which took place in 1997 between more traditional immunologists (Langman) and less ones (Dembic, Coutinho), and by proposing a very simple and illustrative computer simulation allowing a beginning of “formalization” of the self-assertion perspective. I will conclude by discussing the practical impact that such a perspective should have on the conception of “intrusion detectors” for vulnerable systems such as computers, and why a growing number of immunologists, like Varela twenty years ago, plead for going beyond this too narrow vision of immune system as “intrusions detector” to rather privilege its “homeostatic character”.

Innate immunity now occupies a central role in immunology. However, artificial immune system models have largely been inspired by adaptive not innate immunity. This paper reviews the biological principles and properties of innate immunity and, adopting a conceptual framework, asks how these can be incorporated into artificial models. The aim is to outline a meta-framework for models of innate immunity.

In this conceptual paper, we consider the state of artificial immune system (AIS) design today, and the nature of the immune theories on which they are based. We highlight the disagreement amongst many immunologists regarding the concept of self–non-self discriminations in the immune system, and go on describe on such model that removes altogether the requirement for self–non-self discrimination. We then identify the possible inspiration ideas for AIS that can be gained from such new, and often radical, models of the immune system. Next, we outline a possible approach to designing AIS that are inspired by new immune theories, following a suitable methodology and selecting appropriate modelling tools. Lastly, we follow our approach and present an example of how the AIS designer might take inspiration from a specific property of a new immune theory. This example highlights our proposed method for inspiring the design of the next generation of AIS.

An immune system without tissue is like evolution without genes. Something very important is missing. Here we present the novel concept of tissue for artificial immune systems. Much like the genetic representation of genetic algorithms, tissue provides an interface between problem and immune algorithm. Two tissue-growing algorithms are presented with experimental results illustrating their abilities to dynamically cluster data and provide useful signals. The use of tissue to provide an innate immune response driving the adaptive response of conventional immune algorithms is then discussed.

Dendritic cells are antigen presenting cells that provide a vital link between the innate and adaptive immune system. Research into this family of cells has revealed that they perform the role of co-ordinating T-cell based immune responses, both reactive and for generating tolerance. We have derived an algorithm based on the functionality of these cells, and have used the signals and differentiation pathways to build a control mechanism for an artificial immune system. We present our algorithmic details in addition to some preliminary results, where the algorithm was applied for the purpose of anomaly detection. We hope that this algorithm will eventually become the key component within a large, distributed immune system, based on sound immunological concepts.

The role of T-cells within the immune system is to confirm and assess anomalous situations and then either respond to or tolerate the source of the effect. To illustrate how these mechanisms can be harnessed to solve real-world problems, we present the blueprint of a T-cell inspired algorithm for computer security worm detection. We show how the three central T-cell processes, namely T-cell maturation, differentiation and proliferation, naturally map into this domain and further illustrate how such an algorithm fits into a complete immune inspired computer security system and framework.

Administered antibodies can suppress humoral immune response. Though there are two hypotheses explaining the suppression, such as the epitope-masking and Fc-receptor mediated suppression, the epitope-masking hypothesis has garnered more supports. To better understand how the immune suppression works and to gain a quantitative and qualitative insight, we developed the first mathematical immune suppression model based on the epitope-masking hypothesis. However, because the hypothesis does not account for the actual B suppression mechanism, the fact that antigen-depletion induces the arrest of proliferating B cells was incorporated to the model. The model can reproduce immune suppression phenomena and complement the epitope-masking hypothesis by suggesting that the key mechanism for the suppression is the arrest of proliferating B cells and it was shown to be feasible. It is expected that our model gives a new insight to researchers in designing experiments for discovering the underlying mechanism of immune suppression.

Immunological simulations offer the possibility of performing high-throughput experiments

in silico

that can predict, or at least suggest,

in vivo

phenomena. In this paper, we first validate an experimental immunological simulator, developed by the authors, by simulating several theories of immunological memory with known results. We then use the same system to evaluate the predicted effects of a theory of immunological memory. The resulting model has not been explored before in artificial immune systems research, and we compare the simulated

in silico

output with

in vivo

measurements. We conclude that the theory appears valid, but that there are a common set of reasons why simulations are a useful support tool, not conclusive in themselves.

A quaternion model of artificial immune response (AIR) is proposed in this paper. The model abstracts four elements to simulate the process of immune response, namely, antigen, antibody, rules of interaction among antibodies, and the drive algorithm describing how the rules are applied to antibodies. Inspired by the biologic immune system, we design the set of rules as three subsets, namely, the set of clonal selection rules, the set of immunological memory rules, and the set of immunoregulation rules. An example of the drive algorithm is given and a sufficient condition of its convergence is deduced.

Considerable research effort has provided mathematical and computational models of the human immune response under viral infection. However, the quality of simulated results are highly dependent on the choice of modeling strategy. We examine two modeling approaches of HIV pathogenesis: Mathematical and Multi-Agent (or MA) Models. The latter has relatively wider Model Scope due to the agent-rule specification method. Mathematical Models employ Parameter and Population/Subpopulation Level entity granularities with equation-based interaction, while MA Models specify entities at Individual Level, implemented with agents to describe interactions via IF-THEN rules. Compared to the former, MA Models naturally handles entity heterogeneity and spatial non-uniformity, and suffers less from the issue of directly designed dynamics. Both approaches are however, not directly accessible to immunologists due to the need for programming knowledge; hence, closer collaboration between computer scientists and immunologists is necessary.

In this paper, we present a study of the use of an artificial immune system (CLONALG) for solving constrained global optimization problems. As part of this study, we evaluate the performance of the algorithm both with binary encoding and with real-numbers encoding. Additionally, we also evaluate the impact of the mutation operator in the performance of the approach by comparing Cauchy and Gaussian mutations. Finally, we propose a new mutation operator which significantly improves the performance of CLONALG in constrained optimization.

The aim of this work is to propose and validate a new multiobjective optimization algorithm based on the emulation of the immune system behavior. The

rationale

of this work is that the artificial immune system has, in its elementary structure, the main features required by other multiobjective evolutionary algorithms described in literature. The proposed approach is compared with the NSGA2 algorithm, that is representative of the state-of-the-art in multiobjective optimization. Algorithms are tested versus three standard problems (unconstrained and constrained), and comparisons are carried out using three different metrics. Results show that the proposed approach have performances similar or better than those produced by NSGA2, and it can become a valid alternative to standard algorithms.

The (randomized) real-valued negative selection algorithm is an anomaly detection approach, inspired by the negative selection immune system principle. The algorithm was proposed to overcome scaling problems inherent in the hamming shape-space negative selection algorithm. In this paper, we investigate termination behavior of the real-valued negative selection algorithm with variable-sized detectors on an artificial data set. We then undertake an analysis and comparison of the classification performance on the high-dimensional KDD data set of the real-valued negative selection, a real-valued positive selection and statistical anomaly detection techniques. Results reveal that in terms of detection rate, real-valued negative selection with variable-sized detectors is not competitive to statistical anomaly detection techniques on the KDD data set. In addition, we suggest that the termination guarantee of the real-valued negative selection with variable-sized detectors is very sensitive to several parameters.

Despite attempts to legislate them out of existence, spam messages (junk email) continue to fill electronic mailboxes around the world. With spam senders adapting to each technical solution put on the market, adaptive solutions are being incorporated into new products. This paper undertakes an extended examination of the spam-detecting artificial immune system proposed in [1,2], focusing on comparison of scoring schemes, the effect of population size, and the libraries used to create the detectors.

Many algorithms perform data clustering by compressing the original data into a more compact and interpretable representation, which can be more easily inspected for the presence of clusters. This, however, can be a risky alternative, because the simplified representation may contain distortions mainly related to the density information present in the data, which can considerably act on the clustering results. In order to treat this deficiency, this paper proposes an Adaptive Radius Immune Algorithm (ARIA), which is capable of maximally preserving the density information after compression by implementing an antibody adaptive suppression radius that varies inversely with the local density in the space. ARIA is tested with both artificial and real world problems obtaining a better performance than the aiNet algorithm and showing that preserving the density information leads to refined clustering results.

This paper proposes a new immune clonal algorithm, called a quantum-inspired immune clonal algorithm (QICA), which is based on the concept and principles of quantum computing, such as a quantum bit and superposition of states. Like other evolutionary algorithms, QICA is also characterized by the representation of the individual, the evaluation function, and the population dynamics. QICA uses a quantum bit, defined as the smallest unit of information, for the probabilistic representation and a quantum bit individual as a string of quantum bits. In QICA, by quantum mutation operator, we can make full use of the information of the current best individual to perform the next search for speeding up the convergence. Information among the subpopulation is exchanged by adopting the quantum crossover operator for improvement of diversity of the population and avoiding prematurity. We execute the proposed algorithm to solve the benchmark problems with 30,100 and 2000 dimensions and very large numbers of local minima. The result shows that the proposed algorithm can close-to-optimal solution by the less computational cost.

An exact Markov chain model of the B-cell algorithm (BCA) is constructed via a novel possible transit method. The model is used to formulate a proof that the BCA is convergent absolute under a very broad set of conditions. Results from a simple numerical example are presented, we use this to demonstrate how the model can be applied to increase understanding of the performance of the BCA in optimizing function landscapes as well as giving insight into the optimal parameter settings for the BCA.

Helper T(Th) cells regulate immune response by producing various kinds of cytokines in response to antigen stimulation. The regulatory functions of Th cells are promoted by their differentiation into two distinct subsets, Th1 and Th2 cells. Th1 cells are involved in inducing cellular immune response by activating cytotoxic T cells. Th2 cells trigger B cells to produce antibodies, protective proteins used by the immune system to identify and neutralize foreign substances. Because cellular and humoral immune responses have quite different roles in protecting the host from foreign substances, Th cell differentiation is a crucial event in the immune response. The destiny of a naive Th cell is mainly controlled by cytokines such as IL-4, IL-12, and IFN-

γ

. To understand the mechanism of Th cell differentiation, many mathematical models have been proposed. One of the most difficult problems in mathematical modeling is to find appropriate kinetic parameters needed to complete a model. However, it is relatively easy to get qualitative or linguistic knowledge of a model dynamics. To incorporate such knowledge into a model, we propose a novel approach, fuzzy continuous Petri nets extending traditional continuous Petri net by adding new types of places and transitions called fuzzy places and fuzzy transitions. This extension makes it possible to perform fuzzy inference with fuzzy places and fuzzy transitions acting as kinetic parameters and fuzzy inference systems between input and output places, respectively.

This paper describes a multi-agent strategy for blacklisting malicious nodes in a peer-to-peer network that is inspired by the innate immune system, including the recruitment of leukocytes to the site of an infection in the human body. Agents are based on macrophages, T-cells, and tumor necrosis factor, and exist on network nodes that have properties drawn from vascular endothelial tissue. Here I show that this strategy succeeds in blacklisting malicious nodes from the network using non-specific recruitment. This strategy is sensitive to parameters that affect the recruitment of leukocyte agents to malicious nodes. The strategy can eliminate even a large, uniform distribution of malicious nodes in the network.

In most of the existing artificial immune systems, instabilities mainly stem from the empirical pre-definition of a scenario-specific model. In this paper we introduce a self-regulating algorithm into an integrated platform of artificial immune systems based on Model Library. The algorithm can dynamically configure multi-AIS-models according to the “pressure” produced during the course of training and testing, so that the system can automatically adapt to detect various objects. In addition, a novel hybrid evaluation method is proposed to improve the self-adaptability of the system. Experimental results demonstrate that the self-regulating algorithm can achieve better performance as compared with traditional artificial immune systems in terms of false positive and false negative rates.

We have proposed a metaphor “DAnger Susceptible daTa codON” (DASTON) in data subject to processing by Danger Theory (DT) based Artificial Immune System (DAIS). The DASTONs are data chunks or data point sets that actively take part to produce “danger”; here we abstract “danger” as required outcome. To have closer look to the metaphor, this paper furthers biological abstractions for DASTON. Susceptibility of DASTON is important parameter for generating dangerous outcome. In biology, susceptibility of a host to pathogenic activities (potentially dangerous activities) is related to polymorphism. Interestingly, results of experiments conducted for system call DASTONs are in close accordance to biological theory of polymorphism and susceptibility. This shows that computational data (system calls in this case) exhibit biological properties when processed with DT point of view.

The General Suppression Control Framework (GSCF) is a framework inspired by the suppression hypothesis of the immune discrimination theory. The framework consists of five distinct components, the Affinity Evaluator, Cell Differentiator, Cell Reactor, Suppression Modulator, and the Local Environment. These reactive components, each responsible for a specific function, can generate long-term and short-term influences to other components by the use of humoral and cellular signals.

This paper focuses in the design of a control system that aims to balance and navigate a self-balancing robot though obstacles based on the five components in GSCF. The control system demonstrates how simple combination of suppression mechanism can filter and fuses two unstable measurements together to obtain reliable measurement to maintain the balance of a dynamically unstable system. The control system is implemented in a two-wheeled self-balancing robot for its inherited instability can best demonstrate the systems responsiveness to dynamic changes.

Computer generated sounds for music applications have many facets, of which timbre design is of groundbreaking significance. Timbre is a remarkable and rather complex phenomenon that has puzzled researchers for a long time. Actually, the nature of musical signals is not fully understood yet. In this paper, we present a sound synthesis method using an artificial immune network for data clustering, denoted aiNet. Sounds produced by the method are referred to as immunological sounds. Basically, antibody-sounds are generated to recognize a fixed and predefined set of antigen-sounds, thus producing timbral variants with the desired characteristics. The aiNet algorithm provides maintenance of diversity and an adaptive number of resultant antibody-sounds (memory cells), so that the intended aesthetical result is properly achieved by avoiding the formal definition of the timbral attributes. The initial set of antibody-sounds may be randomly generated vectors, sinusoidal waves with random frequency, or a set of loaded waveforms. To evaluate the obtained results we propose an affinity measure based on the average spectral distance from the memory cells to the antigen-sounds. With the validation of the affinity criterion, the experimental procedure is outlined, and the results are depicted and analyzed.

This paper presents an immune-inspired adaptable error detection (AED) framework for Automated Teller Machines (ATMs). This framework two levels, one level is local to a single ATM, while the other is a network-wide adaptable error detection. It employs ideas from vaccination, and adaptability analogies of the immune system. For discriminating between normal and erroneous states, an immune inspired one-class supervised algorithm was employed, which supports continual learning and adaptation. The effectiveness of the local AED was confirmed by its ability of detecting potential failures on an average 3 hours before the actual occurrence. This is an encouraging result in terms of availability, since measures can be devised for reducing the downtime of ATMs.

This work is aimed at presenting a fault detection algorithm composed of multiple interconnected modules, and operating according to the paradigm supported by the danger theory in immunology. This algorithm attempts to achieve significant features that a fault detection system is supposed to have when monitoring a telephone profile system. These features would basically be adaptability due to the strong variation that operational conditions may exhibit over time, and the decrease in the number of false positives, which can be generated when any abnormal behavior is erroneously classified as being a fault. Simulated scenarios have been conceived to validate the proposal, and the obtained results are then analyzed.

Biological immune system is a control system that has strong robusticity and self-adaptability in complex disturbance and indeterminacy environments. The B cell and the antibody in biological immune dynamic process are described in the basic Varela immune network model(BVINM), But the antigen doesn’t exist in this model. An improved Varela immune network model(IVINM) has been presented by appending the antigen in the BVINM in this article. Based on the improved Varela immune network model, An immune controller model is designed and its structure is proposed in the paper. Finally the paper puts forward a simulation example, and analyses the characteristic of the immune controller.

We apply the Clonal Selection principle of the human immune system to solve the Flexible Job-Shop Problem with recirculation. Various practical design issues are addressed in the implemented algorithm, ClonaFLEX; first, an efficient antibody representation which creates only feasible solutions and a bootstrapping antibody initialization method to reduce the search time required. Second, the assignment of suitable mutation rates for antibodies based on their affinity. To this end, a simple yet effective visual method of determining the optimal mutation value is proposed. And third, to prevent premature convergence, a novel way of using elite pools to incubate antibodies is presented. Performance results of ClonaFLEX are obtained against benchmark FJSP instances by Kacem and Brandimarte. On average, ClonaFLEX outperforms a cultural evolutionary algorithm (EA) in 7 out of 12 problem sets, equivalent results for 4 and poorer in 1.

In our previous work, we had been proposed a new artificial immune system named as Attribute Weighted Artificial Immune System (AWAIS) to eliminate the negative effects of taking into account of all attributes in calculating Euclidean distance in shape-space representation which is used in many network-based Artificial Immune Systems (AISs). This system depends on the weighting attributes with respect to their importance degrees in class discrimination. These weights are then used in calculation of Euclidean distances. The performance analyses were conducted in the previous study by using machine learning benchmark datasets. In this study, the performance of AWAIS was investigated for real world problems. The used datasets were medical datasets consisting of Statlog Heart Disease and Pima Indian Diabetes datasets taken from University of California at Irvine (UCI) Machine Learning Repository. Classification accuracies for these datasets were obtained through using 10-fold cross validation method. AWAIS reached 82.59% classification accuracy for Statlog Heart Disease while it obtained a classification accuracy of 75.87% for Pima Indians Diabetes. These results are comparable with other classifiers and give promising performance to AWAIS for that kind of problems.

In this work we propose an immune-based approach for designing of fuzzy systems. From numerical data and with membership function previously defined, the immune algorithm evolves a population of fuzzy classification rules based on the clonal selection, hypermutation and immune network principles. Once AIS are able to find multiple good solutions of the problem, accurate and diverse fuzzy systems are built in a single run. Hence, we construct an ensemble of these classifier in order to achieve better results. An ensemble of classifiers consists of a set of individual classifiers whose outputs are combined when classifying novel patterns. The good performance of an ensemble is strongly dependent of individual accuracy and diversity of its components. We evaluate the proposed methodology through computational experiments on some datasets. The results demonstrate that the performance of the obtained fuzzy systems in isolation is very good. However when we combine these systems, a significant improvement is obtained in the correct classification rate, outperforming the single best classifier.

After a decade of research into the area of Artificial Immune Systems, it is worthwhile to take a step back and reflect on the contributions that the paradigm has brought to the application areas to which it has been applied. Undeniably, there have been a lot of successful stories — however, if the field is to advance in the future and really carve out its own distinctive niche, then it is necessary to be able to illustrate that there are clear benefits to be obtained by applying this paradigm rather than others. This paper attempts to take stock of the application areas that have been tackled in the past, and ask the difficult question “was it worth it ?”. We then attempt to suggest a set of problem features that we believe will allow the true potential of the immunological system to be exploited in computational systems, and define a unique niche for AIS.