Section: New Results
Participants : Anne Auger, Nicolas Bredèche, Philippe Caillou, Cyril Furtlehner, Cécile Germain-Renaud, Cédric Hartland, Nikolaus Hansen, Mohamed Jebalia, Claire Le Baron, Julien Perez, Marc Schoenauer, Michèle Sebag, Xiangliang Zhang.
Several studies pertaining to engineering optimisation (including the work on Structural Design mentioned in Section 6.3.1 ) are on-going, in collaboration with Renault, Dassault and EADS.
Isotherm identification in analytic chromatography, an important challenge for chemical engineers, has been considered within the ACI Chromalgema . A simulation/identification platform had been designed, using Self-Adaptive Evolution Strategies and gradient based methods in a first phase. In a second phase, the use of CMA-ES allowed to significantly improve the results ...and the usability of the platform for chemists [Oops!] .
Multi-Disciplinary Optimisation, a typical application domain for Evolutionary Algorithms since several objectives (usually non-regular) are involved, has been considered in TAO. In 2007, C. LeBaron's PhD (CIFRE Renault, in its 3rd year) is interested in the global optimisation of the propulsion engine, mixing static and dynamic structural optimization with acoustic design.
Along similar lines, TAO participates in the OMD RNTL project, in charge of the general surrogate approach Work Package, and is working on one test-case provided by EADS that concerns the global optimization of a complete launcher considering structural, combustion, and trajectory planification at the same time for several possible missions (satellite positionning around the earth).
Evolutionary Robotics is another domain of application where Machine Learning and Evolutionary Computation can be usefully combined. In TAO, in the recent years, diverse controller representations have been investigated, ranging from classical multi-layer perceptrons (Nicolas Godzik. Une approche évolutionnaire de la robotique modulaire et anticipative . PhD thesis, Université Paris-Sud, September 2005.)to evolved Voronoï-based classifiers (Carlos Kavka. Evolutionary Design of Geometric-Based Fuzzy Systems . PhD thesis, Université Paris-Sud, July 2006.).
During his on-going PhD, C. Hartland first extended the use of an anticipation module to address the reality gap problem (Cedric Hartland and Nicolas Bredèche. Evolutionary robotics, anticipation and the reality gap. In IEEE Intl Conf. on Robotics and Biomimetics (ROBIO) , pp 1640 - 1645, IEEE Press, 2006.). In 2007, the novelty was to consider Echo State Networks (ESNs) [Oops!] . In the scope of learning human demonstrated robot behavior, we showed that by using specific neural network known as Echo State Network, makes it possible to grasp relevant information, both punctual and temporal, regarding the demonstrated sequence. Experiments were conducted on a real world khepera robot and showed promising directions (C. Hartland, N. Bredeche. Using Echo State Networks for Robot Navigation Behavior Acquisition. Accepted at the 4rd IEEE International Conference on Robotics and Biomimetics (ROBIO 2007). Sanya, China, December 2007.). We also considered using ESN in the scope of evolutionary optimisation without oracle [Oops!] (see section 6.3.1 ).
We are also concerned with optimizing robot locomotion and morphologies. Recent works this year focused on bridging the reality gap for an evolved locomotion controler for a tetrapodal walking robot using Central Pattern Generator as a locomotion basic block. On the practical aspects, we implemented a howebrew bios update for a real world robotic kit and then ran an evolved controler on the real tetrapodal robot (in the scope of the SCOUT STIC-asie projet). We also proposed a representation formalism to deal with representing robot morphologies. The goal is to use such a representation to evolve morphologies for a given objective (typicaly a locomotion gait) (Master's thesis M2R, 2007).
Autonomic Computing, acknowledged a Grand Challenge by IBM in 2001 (J.O. Kephart, D.M. Chess, The Vision of Autonomic Computing, in Computer Magazine, reprinted by IEEE Press, pp 41-50, 2003.)aims at self-aware, self-healing and self-configuring complex computer systems.
Autonomic Grids were considered a promising field of ML applications as Cécile Germain-Renaud provided both her expertise and extensive datasets related to the EGEE (Enabling Grids for E-Science in Europe, infrastructure project (2001-2003), (2003-2007), (2008-2010).)grid. A first contribution related to Feature Construction was motivated by the EGEE job modelling application. Exploiting prior knowledge regarding the data heterogeneity (e.g. the job lifecycle depends on both the current grid load, and the job user's expertise), the dataset was aggressively sub-sampled; the bunch of hypotheses extracted from every sub-sample was used along the dual attribute/example clustering principle (Noam Slonim, Naftali Tishby: Document clustering using word clusters via the information bottleneck method. SIGIR 2000: 208-215.)to derive relevant clusters and sidestep the lack of natural metric on the native representation [Oops!] , [Oops!] . Self-protection has been explored through abrupt changepoint detection methods [Oops!] with the perspective of new applications for MAB algorithms. In the area of grid scheduling, preliminary work [Oops!] showed that grid job traffic shares some properties with Internet traffic. Julien Perez's PhD considers reinforcement learning applied to scheduling, with utility function including the QoS requirements of reactive (interactive) grids, in line with previous work on the middleware (C. Germain-Renaud, C. Loomis, J. Moscicki, and R. Texier. Scheduling for responsive grids. Journal of Grid Computing , 2007. To appear.C. Germain-Renaud, C. Loomis, R. Texier, and A. Osorio. Grid scheduling for interactive analysis. In HealthGrid 2006 , volume 120 of Studies in Health Technology and Informatics , pp 25–33, 2006. IOS Press.).
The application of EC and ML to software testing has been investigated along the EvoTest STREP (Tanja Voss, ITI, Valencia, Spain coordinator, 2006-2009), where TAO is in charge of the Evolutionary Engine at the core of the search process. The generation of test data is set as an optimisation problem. Depending on the context, the fitness can range from coverage measures (for structural testing), to CPU time or memory consumption (for functional testing of real-time embedded systems). The automatic generation of the Evolutionary Algorithm from the test objectives will be based on GUIDE [Oops!] (see section 5.2 ), and one of the main challenges as far as Evolutionary Optimisation is concerned is to make the search engine fully autonomous, relieving the burden of adjusting evolutionary parameters by trials and errors.
Independently, in a joint work with the Software Engineering group in LRI (Marie-Claude Gaudel and Sandrine Gouraud), Nicolas Baskiotis' PhD has proposed a hybrid approach based on ML and stochastic heuristics to overcome the drawbacks of statistical structural software testing (Alain Denise, Marie-Claude Gaudel, Sandrine-Dominique Gouraud: A Generic Method for Statistical Testing. ISSRE 2004: 25-34.)[Oops!] .
Preliminary studies concerning the modeling of speculative bubbles on a financial market under different rationality frameworks (semet:CF04)were extended as Philippe Caillou's and later Cyril Furtlehner's arrivals strengthened TAO's competence in complex systems (TAO also participates in the European Coordinated Action devoted to Complex Systems, ONCE-CS on the executive side). Further studies were devoted to the interaction of social and economic phenomenons, examining how the structure of the social network governs the macro-indicators in a rent-seeking game [Oops!] , [Oops!] , [Oops!] .
Transportation systems provide other examples of social systems where interesting collective phenomena may emerge from local interactions. We have investigated such complex systems from two perspectives. In the first one we proposes a new approach of traffic reconstruction and prediction based on floating car data, by application of distributed algorithms (belief propagation) and ideas inherited from statistical physics [Oops!] , [Oops!] . In the second perspective, the mechanism of jam emergence due to variability of driver behaviours was analysed using probabilistic tools of queueing networks and statistical physics models (exclusion processes, condensation mechanisms and phase transitions) [Oops!] .