Guillaume Hutzler

Automatic tuning of agent-based models using genetic algorithms

When developing multi-agent systems (MAS) or models in the context of agent-based simulation (ABS), the tuning of the model constitutes a crucial step of the design process. Indeed, agent-based models are generally characterized by lots of parameters, which together determine the global dynamics of the system. Moreover, small changes made to a single parameter sometimes lead to a radical modification of the dynamics of the whole system. The development and the parameter setting of an agent-based model can thus become long and tedious if we have no accurate, automatic and systematic strategy to explore this parameter space. Thats the development of such a strategy that we work on suggesting the use of genetic algorithms. The idea is to capture in the fitness function the goal of the design process (efficiency for MAS that realize a given function, realism for agent-based models, etc.) and to make the model automatically evolve in that direction. However the use of genetic algorithms (GA) in the context of ABS brings specific difficulties that we develop in this article, explaining possible solutions and illustrating them on a simple and well-known model: the food-foraging by a colony of ants.

Matrix-bound PAI-1 supports cell blebbing via RhoA/ROCK1 signaling.

The microenvironment of a tumor can influence both the morphology and the behavior of cancer cells which, in turn, can rapidly adapt to environmental changes. Increasing evidence points to the involvement of amoeboid cell migration and thus of cell blebbing in the metastatic process; however, the cues that promote amoeboid cell behavior in physiological and pathological conditions have not yet been clearly identified. Plasminogen Activator Inhibitor type-1 (PAI-1) is found in high amount in the microenvironment of aggressive tumors and is considered as an independent marker of bad prognosis. Here we show by immunoblotting, activity assay and immunofluorescence that, in SW620 human colorectal cancer cells, matrix-associated PAI-1 plays a role in the cell behavior needed for amoeboid migration by maintaining cell blebbing, localizing PDK1 and ROCK1 at the cell membrane and maintaining the RhoA/ROCK1/MLC-P pathway activation. The results obtained by modeling PAI-1 deposition around tumors indicate that matrix-bound PAI-1 is heterogeneously distributed at the tumor periphery and that, at certain spots, the elevated concentrations of matrix-bound PAI-1 needed for cancer cells to undergo the mesenchymal-amoeboid transition can be observed. Matrix-bound PAI-1, as a matricellular protein, could thus represent one of the physiopathological requirements to support metastatic formation.

The Garden of Chances: A Visual Ecosystem

The Garden of Chances is a computer-generated artwork that makes a link between the realworld climate and a virtual garden of abstract, colored shapes. When the artwork is functioning all day long and all year round, the spectator can see the evolution of the climate as the time passes. The software has been developed as a simulation of a real ecosystem and it relies on multi-agent techniques. In this article, the authors present the basic principles of the software and explain how they use it as a tool to explore both art processes and multi-agent issues of emergence and interpretation.

Computing with bacterial constituents, cells and populations: from bioputing to bactoputing

The relevance of biological materials and processes to computing—aliasbioputing—has been explored for decades. These materials include DNA, RNA and proteins, while the processes include transcription, translation, signal transduction and regulation. Recently, the use of bacteria themselves as living computers has been explored but this use generally falls within the classical paradigm of computing. Computer scientists, however, have a variety of problems to which they seek solutions, while microbiologists are having new insights into the problems bacteria are solving and how they are solving them. Here, we envisage that bacteria might be used for new sorts of computing. These could be based on the capacity of bacteria to grow, move and adapt to a myriad different fickle environments both as individuals and as populations of bacteria plus bacteriophage. New principles might be based on the way that bacteria explore phenotype space via hyperstructure dynamics and the fundamental nature of the cell cycle. This computing might even extend to developing a high level language appropriate to using populations of bacteria and bacteriophage. Here, we offer a speculative tour of what we term bactoputing, namely the use of the natural behaviour of bacteria for calculating.

Emotion Modeling for Intelligent Agents - Towards a Unifying Framework

Existing computational models of emotions, although based on different psychological theories, share common properties and may be seen as the different facets of a common emotional process. We thus present our model GRACE - aiming at unifying existing models into a single architecture while preserving the peculiarities of each of them. We also demonstrate the generality of GRACE in emulating the behavior of these existing models.

Accroche-toi au niveau, j'enlève l'échelle: Éléments d'analyse des aspects multiniveaux dans la simulation à base d'agents

Modelling complex systems often implies to consider entities at several levels of organisation and levels of scales. Taking into account these levels, their mutual interactions, and the organizational dynamics at the interface between levels, is a difficult problem, for which the proposed solutions are often linked to a specific disciplinary field or a particular case study. In order to develop a broader methodology for designing multilevel models, we propose an analytical framework of existing approaches, drawn in particular from the study of three examples in biology and geography. We finally show, through an unifying example, how different approaches can be combined.

When a Collective Outcome Triggers a Rare Individual Event: A Mode of Metastatic Process in a Cell Population

A model of early metastatic process is based on the role of the protein PAI-1, which at high enough extracellular concentration promotes the transition of cancer cells to a state prone to migration. This transition is described at the single cell level as a bi-stable switch associated with a subcritical bifurcation. In a multilevel reaction-diffusion scenario, the micro-environment of the tumor is modified by the proliferating cell population so as to push the concentration of PAI-1 above the bifurcation threshold. The formulation in terms of partial differential equations fails to capture spatio-temporal heterogeneity. Cellular-automata and agent-based simulations of cell populations support the hypothesis that a randomly localized accumulation of PAI-1 can arise and trigger the escape of a few isolated cells. Far away from the primary tumor, these cells experience a reverse transition back to a proliferative state and could generate a secondary tumor. The suggested role of PAI-1 in controlling this metastatic cycle is candidate to explain its role in the progression of cancer.

How to gain emotional rewards during human-robot interaction using music? formulation and propositions

In this paper, we present arguments for the need of emotion modelling and we define elements for a study in Human-Robot Interaction (HRI) using music. We also propose an adaptation of our generic model of emotions (GRACE) to give a precise idea of how to design emotional intelligence for a robot with music-related abilities.

Validation of an agent based system using Petri Nets

Hsim is an agent-based simulator used to model the interactions of macromolecules in a 3D virtual cell surrounded by a membrane. This simulator is not dedicated to a specific model of macromolecular interactions but allows the modelling of any kind of interactions, described using a modelling language, written by the user of the simulator and representing reactions between the various molecules, including the formation of complexes.Our aim was to verify the validity of the users model and to exhibit properties of the model without needing to run a simulation. To this end, we construct a Petri Net to determine all the possible macromolecular assemblies and use the Petri Net algebra to verify chosen properties.

Grounding Virtual Worlds in Reality

We suggest in this article a new paradigm for the representation of data, which is best suited for the real-time visualization and sonorisation of complex systems, real or simulated. The basic idea lies in the use of the garden metaphor to represent the dynamic evolution of interacting and organizing entities. In this proposal, multiagent systems are used to map between given complex systems and their garden-like representation, which we call Data Gardens (DG). Once a satisfying mapping has been chosen, the evolution of these Data Gardens is then driven by the real-time arrival of data from the system to represent and by the endogenous reaction of the multiagent system, immersing the user within a visual and sonorous atmosphere from which he can gain an intuitive understanding of the system, without even focusing his attention on it. This can be applied to give life to virtual worlds by grounding them in reality using real world data.