EN FR
EN FR


Section: New Results

Complex Systems

Participants : Jamal Atif, Nicolas Bredèche, Matthias Brendel, Cyril Furtlehner, Philippe Caillou, Jean-Marc Montanier, Hélène Paugam-Moisy, Marc Schoenauer, Michèle Sebag.

Evolutionary AI Planning:

Divide And Evolve (DAE) DAE solves AI-planning problems by using an Evolutionary Algorithm to sequentially divide them into hopefully simpler problems that are handled by some embedded “classical” planner. Within the ANR project DESCARWIN, work has continued in collaboration with Thalès Research & Technology and ONERA Toulouse. A large part of the work this year has been devoted to writing a brand new version of the DAE software, facing difficulties of parallelization [90] . The resulting program entered the 7th International Planning Competition (IPC 2011) at the 21st International Conference on Planning and Scheduling (ICAPS 2011) and won the Gold Medal in the Temporal Track. Note that the Silver Medal was won by Vincent Vidal, also member of the DESCARWIN team, using his planner YAHSP2 – the one that won the Gold Medal while embedded in DAE, thus demonstrating one more the added value of the DAE approach. Meanwhile, because DAE has many parameters (like most Evolutionary Algorithms), parameter tuning within DAE remains a difficult task, and an original approach has been proposed to learn the parameters based on some instance features [49] , [50] , [51] . Note that this method is however relevant of the “Crossing the Chasm” SIG (see Section 6.3 ), as it can be applied to any optimization algorithm that handles several instances of the same class.

Distributed Autonomous Robotics

Resuming work done in 2010, we investigated further the issue of robotic swarm control whenever the environment is partially or completely unknown. This research is at the cross-road of Evolutionary Computation, Machine Learning and Robotics, and a light influence from Evolutionary Ecology, but with a strong focus on engineering (ie. the goal remains to design algorithms). The topic we are interested in is the design of environment-driven self-adaptive distributed algorithms to enable survival at the level of a population of independent robotic units. The population is limited in size, and hardware implementation within real robots has already been achieved [7] . We have also focused our attention on specific aspects of swarm evolutionary dynamics under specific constraints, including the evolution of cooperative and/or altruistic behaviours [53] , [52] . This research yielded interesting results, such as the emergence of altruistic behavior under simple, but specific, algorithmic constraint, as well as tuning mechanism to control the level of altruistic behavior in a population of robots. Perspectives of this work is currently under investigation.

The work done in 2010 about the division of labor among asynchronous and decentralized agents, where each agent is modelled from the competition between two spiking neurons, was further analyzed within a spatio-temporal (simulated) frame. The phase transitions between the asynchronous, the aperiodic and periodic synchronous regimes (depending on the sociability and excitability of the agents) was confirmed, with some counter-intuitive results about the overall merits and efficiency of synchronous behaviors [23] .

We have also explored objective-driven online learning within real robotic hardware, both for single robot online behavior learning [78] as well as small group of robots for pattern formation learning [43] . Our activity in Evolutionary Robotics has also been strenghtened by the publication of book which gather several contributions from major actors in the field [75] , including an introduction paper on current trends and challenges in this domain [76] .

From a slightly different perspective, our work on evolving generative and developmental representations has been continued, with an extensive study of robustness within developmental systems [8] and an investigation of the temporal dynamics at work within genetic regulatory networks for design [32] . While not stricly related to robotics, these contributions share the distributed nature of computation and ultimatly aim at providing an efficient representation for designing and controlling large scale passive or active assembly of units (e.g. robots with complex morphologies).

Additionally, at the crossroad of Machine Learning and Evolutionary Computation, a new Reinforcement Learning approach based on modelling the user's preferences was proposed [12] , [84] ; in the so-called Preference-based Policy Learning, the robot demonstrates some behaviors, is informed of the user's preferences, builds a model of the user's preferences and self-trains to build a new behavior hopefully more satisfactory according to the conjectured user's preferences.

Statistical Physics Perspective

Basic tools from statistical physics (scaling, mean-field techniques and associated distributed algorithms, exactly-solvable models) and probability have been used to model and optimize complex systems, either standalone or combined with MABS approaches. Results are

In the context of the ANR TRAVESTI project dealing with spatial and temporal modelling of traffic congestion we have studied in [83] some specific properties of the Belief Propagation algorithm used for inference; we have proposed a way to encode dependencies between real variables with a latent binary MRF [89] ; we have analyzed macro-states on traffic data and how these relate to belief propagation fixed points [35] .

Also in the ANR Travesti context [36] , [88] for modelling congestion at the microscopic level we have proposed a new family of queueing processes where the service rate is coupled stochastically to the number of clients. With this formulation we have been able to relate an asymmetry between acceleration and braking to some condensation mechanism.In this framework we have also proposed a large deviation formulation of the fundamental diagram of traffic flow.

In the design of multi-objective message passing algorithms we have shown how the state-of-the art MAXSAT solver SP-Y can be used directly by an endogeneous clause elimination procedure, to sample Pareto Front of multi-objective 3-SAT problems [87] .

Multi-agent and games

Within the InnovNation serious game project, in collaboration with Paraschool and BlueNove, we developed an e-brainstorming prototype for collaborative ideation. The prototype was tested on 150 students and is being improved to be commercialized. The realist network generator used to analyze the prototype games was also used to study the labor market, and to study the relative importance of friends and colleagues while seeking for a job without ([44] or with [64] variable information transmission speed. We have shown that friends were the most useful when the labor market was at the equilibrium (approximately the same number of jobs and applicants). To analyze the logs of multi-agent based simulations (for example for the InnovNation project), we developed a tool to describe homogeneous agent clusters and their evolution ([73] ). We use the cluster description to build agent models and generate new simulations with this model to validate the results. Also in the complex social system analysis context, we applied LSA text mining tools on research projects and patent category descriptions to associate research clusters to their main research fields [70] . This was used to build a classification of the French research clusters based on their context, and especially the adequation of the cluster research specialization with the regional specialization [72] .

Image understanding

Within the context of image understanding, a new sequential recognition framework has been proposed in [10] . Sequential image understanding refers to the decision making paradigm where objects in an image are successively segmented/recognized following a predefined strategy. Such an approach generally raises some questions about the ‘‘best’’ segmentation sequence to follow and/or how to avoid error propagation. In [10] , we propose original approaches to answer these questions in the case where the objects to segment/recognize are represented by a model describing the spatial relations between objects. The process is guided by a criterion derived from visual attention, and more precisely from a saliency map, along with some spatial information to focus the attention. This criterion is used to optimize the segmentation sequence. Spatial knowledge is also used to ensure the consistency of the results and to allow backtracking on the segmentation order if needed. The proposed approach was applied for the segmentation of internal brain structures in magnetic resonance images. The results show the relevance of the optimization criteria and the interest of the backtracking procedure to guarantee good and consistent results. From a logical standpoint, sequential object recognition is formulated as an abduction process in [14] , [66] . A scene is viewed as an observation and the task of interpretation is considered as the “best” explanation considering the prior knowledge about the scene context. Towards this aim, we introduce an algebraic-based framework unifying mathematical morphology, description logics and formal concept analysis. We propose to compute the best explanations of an observation through algebraic erosion over the Concept Lattice of a background theory which is efficiently constructed using tools from Formal Concept Analysis. We show that the defined operators are sound and complete and satisfy important rationality postulates of abductive reasoning.