James Decraene

Autonomous Bee Colony Optimization for multi-objective function

An Autonomous Bee Colony Optimization (A-BCO) algorithm for solving multi-objective numerical problems is proposed. In contrast with previous Bee Colony algorithms, A-BCO utilizes a diversity-based performance metric to dynamically assess the archive set. This assessment is employed to adapt the bee colony structures and flying patterns. This self-adaptation feature is introduced to optimize the balance between exploration and exploitation during the search process. Moreover, the total number of search iterations is also determined/optimized by A-BCO, according to user pre-specified conditions, during the search process. We evaluate A-BCO upon numerical benchmark problems and the experimental results demonstrate the effectiveness and robustness of the proposed algorithm when compared with the Non-dominated Sorting Genetic Algorithm II and the latest Multi-objective Bee Colony Algorithm proposed to date.

Evolving agent-based simulations in the clouds

Evolving agent-based simulations enables one to automate the difficult iterative process of modeling complex adaptive systems to exhibit pre-specified/desired behaviors. Nevertheless this emerging technology, combining research advances in agent-based modeling/simulation and evolutionary computation, requires significant computing resources (i.e., high performance computing facilities) to evaluate simulation models across a large search space. Moreover, such experiments are typically conducted in an infrequent fashion and may occur when the computing facilities are not fully available. The user may thus be confronted with a computing budget limiting the use of these “evolvable simulation” techniques. We propose the use of the cloud computing paradigm to address these budget and flexibility issues. To assist this research, we utilize a modular evolutionary framework coined CASE (for complex adaptive system evolver) which is capable of evolving agent-based models using nature-inspired search algorithms. In this paper, we present an adaptation of this framework which supports the cloud computing paradigm. An example evolutionary experiment, which examines a simplified military scenario modeled with the agent-based simulation platform MANA, is presented. This experiment refers to Automated Red Teaming: a vulnerability assessment tool employed by defense analysts to study combat operations (which are regarded here as complex adaptive systems). The experimental results suggest promising research potential in exploiting the cloud computing paradigm to support computing intensive evolvable simulation experiments. Finally, we discuss an additional extension to our cloud computing compliant CASE in which we propose to incorporate a distributed evolutionary approach, e.g., the island-based model to further optimize the evolutionary search.

Evolvable simulations applied to automated red teaming: a preliminary study

We report preliminary studies on evolvable simulations applied to Automated Red Teaming (ART). ART is a vulnerability assessment tool in which agent-based models of simplified military scenarios are repeatedly and automatically generated, executed and varied. Nature-inspired heuristic techniques are utilized to drive the exploration of simulation models to exhibit desired system behaviors. To date, ART investigations have essentially addressed the evolution of a limited fixed set of parameters determining the agents behavior. We propose to extend ART to widen the range of evolvable simulation model parameters. Using this “evolvable simulation” approach, we conduct experiments in which the agents structure is evolved. Specifically, a maritime scenario is examined where the individual trajectories of belligerent vessels are evolved to break Blue. These experiments are conducted using a modular evolutionary framework coined CASE. The results present counter-intuitive outcomes and suggest that evolvable simulation is a promising technique to enhance ART.

Maritime counter-piracy study using agent-based simulations

A maritime counter-piracy scenario is modeled using the agent-based simulation platform MANA. This simulation model is employed to investigate the requirements for defending a large commercial vessel, relying principally on non-lethal deterrents, against pirate hijacking. To assist this research, we utilize the data farming methodology to identify the landscape of possibilities, i.e., data farming is employed to efficiently generate and examine a large range of simulation model variants which altogether depict a comprehensive overview of potential outcomes. Moreover, we complement this study through the evaluation of Automated Red Teaming (ART) to uncover the commercial vessels critical vulnerabilities against pirates. ART differs from data farming by exploiting the principles of artificial evolution to automatically generate simulation model variants of interest. Both data farming and ART techniques are applied to our maritime counter-piracy simulation model in this paper. The experimental results provide complementary insights which may assist defense experts in future analyses and decision makings.

Automated modeling and analysis of agent-based simulations using the CASE framework

We present a modular evolutionary framework, coined CASE for complex adaptive system evolver, to automate the modeling and analysis of agent-based simulations (ABSs). The field of agent-based modeling is rapidly growing due to its capabilities to expose the emerging complex phenomena occurring in a wide range of natural and artificial systems such as biological cells, societies, battlefields, stock markets, etc. Nevertheless, studying agent-based simulations is a complicated, interdisciplinary and time-consuming process. Indeed, a large number of simulation parameters has to be considered to identify and fully understand the conditions leading to the emerging phenomena of interest. To tackle this difficulty, the study of ABSs is thus typically conducted in an iterative manner, where each iteration includes the successive and manual modeling of ABSs and analysis of simulation outcomes. To automate this iterative and time-consuming process, we propose CASE, a platform-independent framework capable of evolving ABSs to exhibit the desired emerging behaviors. Through this evolutionary approach, the examination, i.e., the modeling, execution and analysis, of ABSs is automated. This process automation significantly facilitates the examination of complex systems using ABSs. In this paper, we present in detail this modular evolutionary framework which is illustrated with an example experiment. In this experiment, CASE is utilized for Automated Red Teaming, a simulation-based vulnerability assessment technique commonly employed by defense analysts. The aim of this paper is to introduce this flexible computational framework which may potentially benefit related fields involving agent-based simulations such as the gaming or financial industries.

Research advances in automated red teaming

We present, combine and apply novel research advances to Automated Red Teaming (ART). ART is an automated vulnerability assessment tool which is employed to uncover the hard-to-predict and potentially critical elements of military operations. ART is principally based on the use of agent-based modelling/simulation, data farming and evolutionary computation. In this paper, we present two distinct computational methods to address multiple issues of ART: constraint handling and computing budget. These novel techniques originate from the research fields of evolutionary computation and cloud computing. These techniques are applied to a military toy model which was developed with the agent-based simulation platform MANA. We then discuss another potential bottleneck of ART: many-objective optimization. The aim of this research is to optimize ART to best assist defense experts in operational analysis and, ultimately, in critical decision making.

A Multidisciplinary Survey of Computational Techniques for the Modelling, Simulation and Analysis of Biochemical Networks

All processes of life are controlled by networks of interacting biochemical components. The purpose of modelling these networks is manifold. From a theoret- ical point of view it allows the exploration of network structures and dynamics, to find emergent properties or to explain the organisation and evolution of networks. From a practical point of view, in silico experiments can be performed that would be very expensive or impossible to achieve in the laboratory, such as hypothesis-testing with regards to knock-out experiments or overexpression, or checking the validity of a proposed molecular mechanism. The literature on modelling biochemical networks is growing rapidly and the motivations behind different modelling techniques are some- times quite distant from each other. To clarify the current context, we review several of the most popular methods and outline the strengths and weaknesses of deterministic, stochastic, probabilistic, algebraic and agent-based approaches. We then present a com- parison table which allows one to identify easily key attributes for each approach such as: the granularity of representation or formulation of temporal and spatial behaviour. We describe how through the use of heterogeneous and bridging tools, it is possible to unify and exploit desirable features found in differing modelling techniques. This paper provides a comprehensive survey of the multidisciplinary area of biochemical networks modelling. By increasing the awareness of multiple complementary modelling approaches, we aim at offering a more comprehensive understanding of biochemical networks.

Effective crowd control through adaptive evolution of agent-based simulation models

We report an approach to achieve effective crowd control strategies through adaptively evolving an agent-based model of Crowd Simulation for Military Operations (COSMOS). COSMOS is a complex system simulation platform developed to address challenges posed by the Military Operations in Urban Terrains (MOUT). Modeling and simulating soldiers tactical behaviors in MOUT scenarios is challenging due to the complex and emerging behaviors of crowds and large parameter space of the models. Consequently, it is difficult to search for effective crowd control strategies through tuning the model parameters manually. We employ an adaptive evolutionary computation approach, using the Complex Adaptive Systems Evolver (CASE), to address this challenge. Specifically, we conduct experiments using a “building-protection” scenario, where the operation plans of soldier agents are adaptively evolved to best control a crowd. The results suggest this approach using agent-based simulation and evolutionary computation techniques is promising for the study of complex military operations.

THE EVOLUTION OF COMPLEXITY IN SELF-MAINTAINING CELLULAR INFORMATION PROCESSING NETWORKS

We examine the role of self-maintenance (collective autocatalysis) in the evolution of computational biochemical networks. In primitive proto-cells (lacking separate genetic machinery) self-maintenance is a necessary condition for the direct reproduction and inheritance of what we here term Cellular Information Processing Networks (CIPNs). Indeed, partially reproduced or defective CIPNs may generally lead to malfunctioning or premature death of affected cells. We explore the interaction of this self-maintenance property with the evolution and adaptation of CIPNs capable of distinct information processing abilities. We present an evolutionary simulation platform capable of evolving artificial CIPNs from a bottom-up perspective. This system is an agent-based multi-level selectional Artificial Chemistry (AC) which employs a term rewriting system called the Molecular Classifier System (MCS.bl). The latter is derived from the Holland broadcast language formalism. Using this system, we successfully evolve an artificial CIPN to improve performance on a simple pre-specified information processing task whilst subject to the constraint of continuous self-maintenance. We also describe the evolution of self-maintaining, cross-talking and multi-tasking, CIPNs exhibiting a higher level of topological and functional complexity. This proof of concept aims at contributing to the understanding of the open-ended evolutionary growth of complexity in artificial systems.

Studies on Pareto-based multi-objective competitive coevolutionary dynamics

Competitive coevolutionary algorithms are stochastic population-based search algorithms. To date, most competitive coevolution research has been carried in the domain of single-objective optimization. We propose a novel competitive coevolutionary framework to explore Pareto-based multi-objective competitive coevolution. This framework utilizes the hypervolume indicator and fitness sharing mechanism to address disengagement and over-specialisation issues. A diversity-driven evolutionary selection scheme is utilized to deal with the loss of fitness gradient problem. Several series of experiments are conducted using multi-objective two-sided competitive games. The results suggest that Pareto-optimal solutions can effectively be found using our proposed coevolutionary framework.