Fullscreen
Loading...
 

Print

ISC-PIF Researcher & Fmr. Director  //




René Doursat



Image

René Doursat is a researcher and lecturer in computer science, focusing on complex systems and neural computation. An alumnus of the Ecole Normale Supérieure in Paris, he made a detour of several years through the software industry following his 1991 PhD and initial post-docs, then returned to academia full-time in 2004 as a Visiting Assistant Professor at the University of Nevada. He has held a research position at the Complex Systems Institute in Paris since 2007, and was also its director for two years in 2009-2010 (but has since handed over this day-to-day management responsibility in order to dedicate himself again to full-time research). In 2010, he defended and obtained his Habilitation diploma (ability to supervise graduate research) from Université Pierre et Marie Curie-Paris 6. The main theme of Dr. Doursat’s research is the computational modeling and simulation of complex multi-agent systems, in particular biological and techno-social, which can also inspire novel principles in intelligent systems design. He is especially interested in "self-made puzzles", i.e., the self-organization of complex, articulated morphologies from a swarm of heterogeneous agents, through dynamical, developmental, and evolutionary processes. For example, these emergent patterns can be innovative structures in multicellular organisms, autonomic networks of computing devices, or "mental representations" and imagery made of correlated spiking neurons


See his website here:  http://doursat.free.fr
 

e-mail: doursat (at ) yahoo (dot) com
 

Research


Exploding growth in hardware, software and networks forces us to rethink systems engineering in terms of “meta-designing” mechanisms that would allow those systems to self-assemble, self-regulate and evolve. Toward this goal, understanding natural complex systems, in particular biological (developmental, neural, evolutionary) and social (large-scale networks), could help create a new generation of artificial systems with properties still largely absent from traditional engineering: decentralization, self-organization, and adaptation.

However, in order to make complex systems a source of inspiration for emerging technologies, the most important challenge is not simply to allow self-organization to happen, but first and foremost to guide it. Past a certain fascination for spontaneous order in “statistical” systems, such as random pattern formation or collective motion, another critical issue concerns the reintroduction of programmability and reproducibility into self-organization, i.e., the attempt to re-engineer emergence. More relevant models can thus be found in “morphological” complex systems, such as biological development and social structures. They are composed of sophisticate elements able to combine in various ways to form precise and reproducible architectures.

My work currently addresses two domains: (a) neural computation, where a central question is to understand how the structured symbolic level of cognition (AI) can arise from the underlying complex dynamical system of the brain (neural networks); and (b) artificial life, preoccupied with explaining how complexity and fitness can spontaneously develop and evolve without the need for a higher symbolic level.


Artificial Life: Biologically Inspired Engineering


Image

I am interested in the modeling and simulation of the fundamental principles of self-patterning and self-assembly during embryonic development. The spontaneous making of an entire organism from a single cell is the epitome of a self-organizing and programmable complex system. Through a precise spatiotemporal interplay of genetic switches and chemical gradients, an elaborate form is created without explicit architectural plan or engineering intervention.






Image

I designed original studies involving multi-agent simulations exploiting these fundamental morphogenetic mechanisms. The 2-D model I propose can be construed as (a) moving cellular automata, in which cell rearrangement is influenced by the pattern they form, or (b) heterogeneous collective motion, in which swarm agents differentiate into patterns according to their location. Generalizing this artificial development study to N-D networks, I also looked at the self-assembly of complex but precise network topologies by programmed attachment. Nodes execute the same program in parallel, communicate and differentiate, while links are dynamically created and removed based on “ports” and “gradients” that guide nodes to specific locations.


Neural Dynamics: Large-Scale Spiking Neural Networks


The foundational thesis of cognitive science is that the mind relies on internal states, or representations, that correspond to states of the external world. It operates by creating, assembling and transforming these states, both under the influence of external stimuli and the constraints of its internal dynamics. The nature and structure of these mental states is still an open problem, in particular their embodiment in the neural code (i.e., the laws of organization of the electrophysiological signals).

Cognitive science is currently a federation of disciplines—psychology, AI, linguistics, logic, neuroscience, neural modeling, robotics, etc.—lacking a unified theory. Cognitive models are broadly divided between a logical paradigm, or “cognitivism” based on high-level symbols and formal grammars, and a dynamical paradigm, or “connectionism” based on neural networks and low-level activation equations. Bridging the gap between both approaches requires an intermediate or mesoscopic level of description. By analogy with the discovery of biological macromolecules, which linked macroscopic organisms to microscopic atoms, a new intermediate discipline of “molecular cognition” is needed to explain the laws of perception and language on the basis of elementary neuronal activities.

What could then be the candidate “molecules” of this new Mind-Brain Modern Synthesis? Generally, complex spatiotemporal phenomena in large-scale neural populations have the potential to support the sought-after mesostructure of symbolic and combinatorial systems. Different spiking neural models have focused on different classes of neuronal dynamics at varying levels of biological detail: conductance-based, integrate & fire, pulsed, oscillatory, excitable, rate-coded, binary, etc. They have also explored different forms of temporal order binding these neurons together: synchronization, phase locking, delayed correlations, waves, rhythms, induction, resonance, etc. My projects follow a few of these mesoscopic paradigms, addressing different topics and challenges in robotics, machine vision, linguistic and pattern recognition.
 

Image


Selected publications


Bienenstock, E. & Doursat, R. (1994) A shape-recognition model using dynamical links. Network: Computation in Neural Systems 5(2): 241-258.

Doursat, R. (2008b) Organically grown architectures: Creating decentralized, autonomous systems by embryomorphic engineering. In Organic Computing, R. P. Würtz, ed. Springer-Verlag, pp. 167-200.

Doursat, R. (2008d) Programmable architectures that are complex and self-organized: From morphogenesis to engineering. 11th International Conference on the Simulation and Synthesis of Living Systems (ALIFE XI), August 5-8, 2008, University of Southampton, Winchester, UK. In Artificial Life XI, S. Bullock, J. Noble, R. Watson & M. A. Bedau, eds. MIT Press, pp. 181-188.

Doursat, R. (2006a/2008f) The growing canvas of biological development: Multiscale pattern generation on an expanding lattice of gene regulatory networks. Unifying Themes in Complex Systems Vol VI: Proceedings of the 6th International Conference on Complex Systems, A. A. Minai, D. Braha, Y. Bar-Yam, eds. Springer-Verlag. This volume selected 77 papers from over 300 presented at the ICCS 2006 conference.

Doursat, R. & Bienenstock, E. (2006b) Neocortical self-structuration as a basis for learning. 5th International Conference on Development and Learning (ICDL 2006), May 31-June 3, 2006, Indiana University, Bloomington, IN.

Doursat, R. & Petitot, J. (2005b) Dynamical systems and cognitive linguistics: Toward an active morphodynamical semantics. Neural Networks 18: 628-638. Selected for this special issue among less than 10% of the papers accepted at the IJCNN 2005 conference.

Doursat, R. & Ulieru, M. (2009) TBA. In Special Issue on Engineering Cyber-Physical Ecosystems, IEEE Transactions on Systems, Man and Cybernetics. Accepted.

Geman, S., Bienenstock, E. & Doursat, R. (1992) Neural networks and the bias/variance dilemma. Neural Computation 4: 1-58. Cited over 1500 times (Google Scholar).

Hoelzer, G., Drewes, R., Meier, J. & Doursat, R. (2008) Isolation-by-distance and outbreeding depression are sufficient to drive parapatric speciation in the absence of environmental influences. PLoS Computational Biology 4(7): e1000126 [[doi:10.1371/journal.pcbi.1000126]].

Petitot, J. & Doursat, R. (2009) Cognitive Morphodynamics: Dynamical Morphological Models for Constituency in Perception and Syntax. To appear.





<< back to Researcher's page

Contributors to this page: Rene Doursat and webmaster .
Page last modified on Monday 04 April, 2011 12:16:39 by Rene Doursat.