Search

This paper describes a possible application of swarm intelligence techniques in e-learning: an auditing tool for pedagogical planning. Swarm intelligence techniques can be applied to a web system thanks to the fact that the available online material can be organized in a graph by means of hyperlinks. In this case, the swarm that moves on the graph is composed of students who unconsciously leave pheromones in the environment depending on their success or failure. The paper presents the system and shows its capacity to serve as an auditing tool for courses designed by a pedagogical team.

In this paper, we study quasi-stationarity for a large class of Kolmogorov diffusions. The main novelty here is that we allow the drift to go to $- \infty$ at the origin, and the diffusion to have an entrance boundary at $+\infty$. These diffusions arise as images, by a deterministic map, of generalized Feller diffusions, which themselves are obtained as limits of rescaled birth--death processes. Generalized Feller diffusions take nonnegative values and are absorbed at zero in finite time with probability $1$. An important example is the logistic Feller diffusion. We give sufficient conditions on the drift near $0$ and near $+ \infty$ for the existence of quasi-stationary distributions, as well as rate of convergence in the Yaglom limit and existence of the $Q$-process. We also show that under these conditions, there is exactly one quasi-stationary distribution, and that this distribution attracts all initial distributions under the conditional evolution, if and only if $+\infty$ is an entrance boundary. In particular this gives a sufficient condition for the uniqueness of quasi-stationary distributions. In the proofs spectral theory plays an important role on $L^2$ of the reference measure for the killed process.

We consider the asymptotic behavior of the solution of one dimensional stochastic differential equations and Langevin equations in periodic backgrounds with zero average. We prove that in several such models, there is generically a non-vanishing asymptotic velocity, despite of the fact that the average of the background is zero.

International audience; Cochlear implants are devices that become more and more sophisticated and adapted to the need of patients, but at the same time they become more and more difficult to parameterize. After a deaf patient has been surgically implanted, a specialised medical practitioner has to spend hours during months to precisely fit the implant to the patient. This process is a complex one implying two intertwined tasks: the practitioner has to tune the parameters of the device (optimisation) while the patient's brain needs to adapt to the new data he receives (learning). This paper presents a study that intends to make the implant more adaptable to environment (auditive ecology) and to simplify the process of fitting. Real experiments on volunteer implanted patients are presented, that show the efficiency of interactive evolution for this purpose. This work has partially been funded by the French ANR - RNTS HEVEA project 04T550.

We consider dynamical systems for which the spatial extension plays an important role. For these systems, the notions of attractor, epsilon-entropy and topological entropy per unit time and volume have been introduced previously. In this paper we use the notion of Kolmogorov complexity to introduce, for extended dynamical systems, a notion of complexity per unit time and volume which plays the same role as the metric entropy for classical dynamical systems. We introduce this notion as an almost sure limit on orbits of the system. Moreover we prove a kind of variational principle for this complexity., Comment: 29 pages

Cochlear implants are electronic devices that stimulate directly the auditory nerve to allow totally deaf patients to hear again. This paper presents an interactive evolutionary algorithm (IEA) designed to help finding the best parameters of a cochlear implant for a specific patient.If early cochlear implants only featured one electrode, modern devices now offer up to 22 electrodes, with the hope to be able to transmit more details and help the patient hear better. The work presented in this paper shows however that having more electrodes is not necessarily better.Tests on a patient show surprisingly that some combinations of electrodes yield better results than others, with the problem that there is no real way to determine which electrode is beneficial to speech understanding and which is not.The best result obtained by the patient on a speech understanding evaluation protocol was 48.5/100 after 10 years of fitting sessions by an expert practitioner. For many reasons explained in this paper, the evaluation of the best parameter setting found by the IEA in one day was 91.5/100.

In this paper, we prove an inequality, which we call "Devroye inequality", for a large class of non-uniformly hyperbolic dynamical systems (M,f). This class, introduced by L.-S. Young, includes families of piece-wise hyperbolic maps (Lozi-like maps), scattering billiards (e.g., planar Lorentz gas), unimodal and H{\'e}non-like maps. Devroye inequality provides an upper bound for the variance of observables of the form K(x,f(x),...,f^{n-1}(x)), where K is any separately Holder continuous function of n variables. In particular, we can deal with observables which are not Birkhoff averages. We will show in \cite{CCS} some applications of Devroye inequality to statistical properties of this class of dynamical systems., Comment: Corrected version; To appear in Nonlinearity

This paper describes experiments aimed at adapting Ant Colony Optimization (ACO) techniques to an e-learning environment, thanks to the fact that the available on-line material can be organized in a graph by means of hyperlinks between educational topics. The structure of this graph is to be optimized in order to facilitate the learning process for students.ACO is based on an ant-hill metaphor. In this case, however, the agents that move on the graph are students who unconsciously leave pheromones in the environment depending on their success or failure. In the paper, the whole process is therefore referred to as a "man-hill."Compared to the [13, 14] papers that were providing guidelines for this problem, real-size tests have been performed, showing that man-hills behave differently from ant-hills. The notion of pheromone erosion (rather than evaporation) is introduced.

In a previous SAC-COMPAHEC paper[1], a method was presented using an interactive evolutionary algorithm for cochlear implants fitting.The method has recently been put to test, with very unexpected and encouraging results: in a few words, it seems that the algorithm is capable to obtain much better results than an expert practitioner in many cases.The solutions proposed by the algorithm are counter-intuitive, yet they improve speech recognition drastically. If these preliminary results are confirmed by many more cases, it could mean that experts have been deterministically tuning cochlear implants the wrong way for many years.However, it seems that the very good results obtained by the algorithm depend a lot on the acoustic environment in which the fitting is performed. A broader fitting scheme has therefore been implemented that should overcome this problem, by allowing the patient to sample typical background noises for which the prosthesis should be specifically tuned.In the future, a piece of software will be added to the cochlear implant signal processor that will automatically choose the best setting depending on the kind of sound environment picked up by the microphone.

Content: Part I : Theory P. Collet A SHORT ERGODIC THEORY REFRESHER M. Courbage Notes on Spectral Theory, Mixing and Transport V. Affraimovich, L. Glebsky: Complexity, Fractal Dimensions and Topological Entropy in Dynamical Systems G.M. Zaslavsky, V. Afraimovich: WORKING WITH COMPLEXITY FUNCTIONS G. Gallavotti SRB distribution for Anosov maps P. Gaspard DYNAMICAL SYSTEMS THEORY OF IRREVERSIBILITY W.T. Strunz ASPECTS OF OPEN QUANTUM SYSTEM DYNAMICS E. Shlizerman, V. R. Kedar ENERGY SURFACES AND HIERARCHIES OF BIFURCATIONS. M. Combescure Phase-Space Semiclassical Analysis.Around Semiclassical Trace Formulae Part II : Applications A. Kaplan et al ATOM-OPTICS BILLIARDS F. Family et al CONTROL OF CHAOS AND SEPARATION OF PARTICLES IN INERTIA RATCHETS F. Bardou FRACTAL TIME RANDOM WALK AND SUBRECOIL LASER COOLING CONSIDERED AS RENEWAL PROCESSES WITH INFINITE MEAN WAITING TIMES X. Leoncini et al ANOMALOUS TRANSPORT IN TWO-DIMENSIONAL PLASMA TURBULENCE E. Ott et al THE ONSET OF SYNCHRONISM IN GLOBALLY COUPLED ENSEMBLES OF CHAOTIC AND PERIODIC DYNAMICAL UNITS A.Iomin, G.M. Zaslavsky QUANTUM BREAKING TIME FOR CHAOTIC SYSTEMS WITH PHASE SPACE STRUCTURES S.V.Prants HAMILTONIAN CHAOS AND FRACTALS IN CAVITY QUANTUM ELECTRODYNAMICS M. Cencini et al INERT AND REACTING TRANSPORT M. A. Zaks ANOMALOUS TRANSPORT IN STEADY PLANE FLOWS OF VISCOUS FLUIDS J. Le Sommer, V. Zeitlin TRACER TRANSPORT DURING THE GEOSTROPHIC ADJUSTMENT IN THE EQUATORIALOCEAN A. Ponno THE FERMI-PASTA-ULAM PROBLEM IN THE THERMODYNAMIC LIMIT

An attempt is made to apply ant colony optimization (ACO) heuristics to an E-learning problem: the pedagogic material of an online teaching Web site for high school students is modelled as a navigation graph where nodes are exercises or lessons and arcs are hypertext links. The arcs' valuation, representing the pedagogic structure and conditioning the Web site's presentation, is gradually modified through the release and evaporation of virtual pheromones that reflect the successes and failures of students roaming around the graph. A compromise is expected to emerge between the pedagogic structure as originally dictated by professors, the collective experience of the whole pool of students and the particularities of each individual. The purpose of this study conducted for Paraschool, the leading French E-learning company is twofold: enhancing the Web site by making its presentation intelligently dynamic and providing the pedagogical team with a refined auditing tool that could help it identify the strengths and weaknesses of its pedagogic choices.

This paper presents an optimisation algorithm designed to perform in-situ automatic fitting of cochlear implants.All patients are different, which means that cochlear parametrisation is a difficult and long task, with results ranging from perfect blind speech recognition to patients who cannot make anything out of their implant and just turn it off.The proposed method combines evolutionary algorithms and medical expertise to achieve autonomous interactive fitting through a Personal Digital Assistant (PDA).

In Twenty Problems in the Theory of Cellular Automata, Stephen Wolfram asks “how common computational universality and undecidability [are] in cellular automata.” This papers provides elements of answer, as it describes how another universal cellular automaton than the Game of Life (Life) was sought and found using evolutionary algorithms. This paper includes a demonstration that consists in showing that the presented R automaton can both implement any logic circuit (logic universality) and a simulation of Life (universality in the Turing sense).

This paper is a follow-up of Maarten Keijzer’s award-winning EUROGP’03 paper [Keij03], that suggests using Interval Arithmetic (IA) and Linear Scaling (LS) in Genetic Programming algorithms. The ideas exposed in this paper were so nice that it was decided to experiment with them on a real-world problem on which the LIL research team had some experience and results with: the Ocean Color Inverse Problem.

We study a few problems related to Markov processes of flipping triangulations of the sphere. We show that these processes are ergodic and mixing, but find a natural example which does not satisfy detailed balance. In this example, the expected distribution of the degrees of the nodes seems to follow the power law $d^{-4}$.

The aim of this paper is to show that the common belief, in the evolutionary community, that evaluation time usually takes over 90% of the total time, is far from being always true. In fact, many real-world applications showed a much lower percentage. This raises several questions, one of them being the balance between fitness and operators computational complexity: what is the use of elaborating smart evolutionary operators to reduce the number of evaluations if as a result, the total computation time is increased?.

Nowadays, large medical databases consist of a collection of smaller databases, each on possibly different fields and using different formats, making it increasingly difficult to retrieve valuable information among the thousands of documents retrieved by a simple query. A new Evolutionary Learning Interactive Search Engine (ELISE) feeds on previous user requests to retrieve "alternative" documents that may not be returned by more conventional search engines, in a way that may recall "lateral thinking." Tests on the "Cystic Fibrosis Database" benchmark [1] prove that, while suggesting original documents by adaptation of its internal rules to the context of the user, ELISE is able to improve its recall rate.

We have developed a new method for separating dispersive waves in elastic and viscoelastic rods from three-point strain and one-point velocity measurements. Stress, strain, displacement and velocity, which were supposed to be homogenous in a rod section, were therefore determined at any point of the bar. This method is based on the assumption of a one dimensional and single mode dispersive wave propagation. This method was used in the frequency domain and took into account wave dispersion. It was shown to be stable with respect to noise. The measuring time is increased considerably. Subsequently, the maximum strain which can be measured in a material tested with a classical SHPB (Split Hopinson Pressure Bar) set-up is also increased and is no more limited by the length of the bars. The method was used to evaluate the rate sensitivity of a luminum honecomb over a wide range of strain-rates (from 10 -2 to 1000 s -1 ).

We prove ratio limit theorems (Davies conjecture) for the heat kernel in unbounded domains with Dirichlet boundary conditions under some accessibility assumptions on the boundary. In the case of Benedicks domains when the cone of positive harmonic functions which tend to zero at the boundary is nontrivial we derive the explicit asymptotic for the case with compact holes. We refer to the original papers by S. Martinez, J. San Martin and the author for the general case.

We present in this paper the design of ELISE, an interactive GP system for document retrieval tasks in very large medical databases. The components of ELISE have been tailored in order to produce a system that is capable of suggesting documents related to the query that may be of interest to the user, thanks to evolved profiling information. Tests on the "Cystic Fibrosis Database" benchmark [2] show that, while suggesting original documents by adaptation of its internal rules to the context of the user, ELISE is able to improve its recall rate.

Artificial evolution can be applied to machine learning thanks to genetic programming, for instance, which is a very successful branch of Evolutionary Computing. One would therefore think that by transitivity, because data-mining is one of the main applications of machine learning, artificial evolution could be successful in solving data-mining problems. Whether this is the case or not, it seems that artificial evolution is not much used in data-mining, even though many papers show that EC can provide interesting alternative solutions to standard machine learning approaches. However, a secret weapon of EC is that it is intrinsically parallel. This very special feature allows EC to exploit multi-core computers as well as the massive parallelism of GPGPU cards, with their hundreds of cores. Then, the island model makes it possible to obtain scalable linear (an even in some cases supra-linear) speedup when several machines are used to solve the same problem, therefore allowing one to efficiently execute EC algorithms on massively parallel machines and super-computers. These architectures were quite rare in the past, but are now widely available as parallelism is the path followed by hardware manufacturers to increase the computing power of their processors. Therefore, the future of evolutionary computation is quite bright, and any new developments in EC could probably overtake more standard methods if these methods are sequential by nature, or do not parallelize well. This is a reason why it is essential to encourage researchers to tackle problems in new domains where EC may not yet be a major actor.

International audience; In this paper, we are interested in separating waves in elastic and viscoelastic rods propagating in opposite direction. N strain and P velocity measurements are taken into account. This application of the likelihood method gives a solution in the frequency domain. Using the inverse Fourier transform, one can recover the strain, stress, displacement and velocity at any section of the rod. In experimental conditions, the results are stable against noise when N+P>2 and NP ≠ 0.

Using a new inequality relating the heat kernel and the probability of survival, we prove asymptotic ratio limit theorems for the heat kernel (and survival probability) in general Benedicks domains. In particular, the dimension of the cone of positive harmonic measures with Dirichlet boundary condition can be derived from the rate of convergence to zero of the heat kernel (or the survival probability)., 14 pages

The EASEA language (EAsy Specification of Evolutionary Algorithms) was created in order to allow scientists to concentrate on evolutionary algorithm design rather than implementation. EASEA currently supports two C++ libraries (GALib and EO) and a JAVA library for the DREAM. The aim of this paper is to assess the quality of EASEA-generated code through an extensive test procedure comparing the implementation for EO and GALib of the same test functions.

This paper describes the recently released DREAM (Distributed Resource Evolutionary Algorithm Machine) framework for the automatic distribution of evolutionary algorithm (EA) processing through a virtual machine built from large numbers of individual machines linked by standard Internet protocols. The framework allows five different user entry points which depend on the knowledge and requirements of the user. At the highest level, users may specify and run distributed EAs simply by manipulating graphical displays. At the lowest level the framework turns becomes a P2P (Peer to Peer) mobile agent system, that may be used for the automatic distribution of a class of processes including, but not limited to, EAs.

This paper describes how animat-based "food foraging" techniques may be applied to the design of low-level image processing algorithms. First, we show how we implemented the food foraging application using the EASEA software package. We then use this technique to evolve an animat and learn how to move inside images and detect high-gradient lines with a minimum exploration time. The resulting animats do not use standard "scanning + filtering" techniques but develop other image exploration strategies close to contour tracking. Experimental results on grey level images are presented.

Evolutionary algorithms are not straightforward to implement and the lack of any specialised language forces users to reinvent the wheel every time they want to write a new program. Over the last years, evolutionary libraries have appeared, trying to reduce the amount of work involved in writing such algorithms from scratch, by offering standard engines, strategies and tools. Unfortunately, most of these libraries are quite complex to use, and imply a deep knowledge of object programming and C++. To further reduce the amount of work needed to implement a new algorithm, without however throwing down the drain all the man-years already spent in the development of such libraries, we have designed EASEA (acronym for EAsy Specification of Evolutionciry Algorithms): a new high-level language dedicated to the specification of evolutionary algorithms. EASEA compiles .ez files into source files in a target language, containing function calls to a chosen existing library. The resulting source file is in turn compiled and linked with the library to produce an executable file implementing the evolutionary algorithm specified in the original .ez file. EASEA vO.4 is available at: http://www-rocq.inria.fr/EVO-Lab/.

Non-linear Iterated Function Systems (IFSs) are very powerful mathematical objects related to fractal theory, that can be used in order to generate (or model) very irregular shapes. We investigate in this paper how Genetic Programming techniques can be efficiently exploited in order to generate randomly or interactively artistic "fractal" 2D shapes. Two applications are presented for different types of non-linear IFSs: # interactive generation of Mixed IFSs attractors using a classical GP scheme, # random generation of Polar IFSs attractors based on an "individual" approach of GP.

In this paper we consider the class of stochastic stationary sources induced by one-dimensional Gibbs states, with Holder continuous potentials. We show that the time elapsed before the source repeats its first n symbols, when suitably renormalized, converges in law either to a log-normal distribution or to a finite mixture of exponential random variables. In the first case we also prove a large deviation result.

We study the set of solutions of the complex Ginzburg-Landau equation in $\real^d, d\OO(\log(1/\epsilon))$, Comment: 24 pages

Let T be an expanding transformation defined on A —(J A{, i= 1being a finite collection of connected open bounded subsets of 2Rn,such that T Acontains strictly Aand Tis Markovian. We prove the existence of a quasi-stationary distrition for T. We show that the T-invariant probability on the limit Cantor set is Gibbsian with potential Log|_DT|. Using the Hilbert projective metric we prove that both distributions are weak limits of conditional laws of probabilities, the speed of convergence being exponential. These results develop a previous work by G. Pianigiani and J.A. Yorke.

Given two pointsx, y∈S 1 randomly chosen independently by a mixing absolutely continuous invariant measure μ of a piecewise expanding and smooth mapf of the circle, we consider for each e>0 the point process obtained by recording the timesn>0 such that |f n (x)−f n (y)|≦e. With the further assumption that the density of μ is bounded away from zero, we show that when e tends to zero the above point process scaled by e−1 converges in law to a marked Poisson point process with constant parameter measure. This parameter measure is given explicity by an average on the rate of expansion off.

[EN] Evolutionary algorithms have proved to be efficient for solving complicated optimization problems. On the other hand, the many-core architecture in graphical cards "General Purpose Graphic Processing Unit" (GPGPU) offers one of the most attractive cost/performance ratio. Using such hardware, the manuscript shows how an efficiently implemented genetic algorithm with a simple fitness function allows boosting the determination of zeolite structures. A case study is presented. © 2011 the Societies Owner.

We investigate the existence of a global semiflow for the complex Ginzburg-Landau equation on the space of bounded functions in unbounded domain. This semiflow is proven to exist in dimension 1 and 2 for any parameter values of the standard cubic Ginzburg-Landau equation. In dimension 3 we need some restrictions on the parameters but cover nevertheless some part of the Benjamin-Feijer unstable domain., uuencoded dvi file (email: collet@orphee.polytechnique.fr)

New bounds are given for the L2-norm of the solution of the Kuramoto-Sivashinsky equation $$\partial _t U(x,t) = - (\partial _x^2 + \partial _x^4 )U(x,t) - U(x,t)\partial _x U(x,t)$$ , for initial data which are periodic with periodL. There is no requirement on the antisymmetry of the initial data. The result is $$\mathop {\lim \sup }\limits_{t \to \infty } \left\| {U( \cdot ,t)} \right\|_2 \leqslant const. L^{8/5} $$ .

The local singularities for a class of random measures, obtained by random iterated multiplications, are investigated using the thermodynamic formalism. This analysis can be interpreted as a rigorous study of the phase transition of a system with random interactions.

In the context of interacting particle systems, we study the influence of the action of the semigroup on the concentration property of Lipschitz functions. As an application, this gives a new approach to estimate the relaxation speed to equilibrium of interacting particle systems. We illustrate our approach in a variety of examples for which we obtain several new results with short and non-technical proofs. These examples include the symmetric and asymmetric exclusion process and high-temperature spin-flip dynamics ("Glauber dynamics"). We also give a new proof of the Poincar\'e inequality, based on coupling, in the context of one-dimensional Gibbs measures. In particular, we cover the case of polynomially decaying potentials, where the log-Sobolev inequality does not hold., Comment: 33 pages, J. Math. Phys. 49 (2008). A typo in inequality (24) was corrected