Connectionists: postdoctoral fellowship proposal

[Bernard Girau]

A postdoctoral position is available at INRIA/Lorraine, Nancy, France.
The deadline for submitting applications is very close: 14 of July 2006.


------------------------------------------------------------------------ 
-------


               INRIA-LORRAINE

        INRIA 2006 POSTDOCTORAL FELLOWSHIP PROPOSAL



------------------------------------------------------------------------ 
-------


City :
Nancy, France

Title of Research Project:
Data-structured connectionist models for visual and multimodal  perception

INRIA team in which the research would be effected:
Cortex

Fellowship supervisor:
Bernard Girau, Assistant professor at University Nancy 2, Cortex project


Description of Research:

Our connectionist research group carries out numerous projects about
the use of neural networks for autonomous systems (autonomous robotics,
embedded processing of physiological signals, ... etc). These works
take advantage of the elementary and massively distributed computations
of connectionist models. They include for example perceptive
pre-processing modules (vision, olfaction) or procedural control, that
aim at interacting autonomously in a real-time and embedded way.

The connectionist paradigm has to be the foundation of the autonomy
of the processings we develop. Therefore their specific very fine-grain
massive parallelism has to be fully exploited, where computation units
take place in a very dense information stream. The organisation and
the interpretation of such a stream is not a simple problem. To solve
it, our research group takes advantage of two approaches, so as to
study the connectionist models and algorithms from a functional point
of view (bio-inspiration) as well as from a computational point of
view (properties of the corresponding computation models). This
subject follows both approaches.

Standard connectionist models for visual perception usually consist
of associated modules, each of them being retinotopically structured.
In such models, images are the inputs of the first module or layer
of neurons [1]. We study an alternate approach where the neural
architecture mimics the natural structure of data (2-dimensional
for visual perception), so that these data directly configure and
parameterize the neural resources. In such an approach, the
retinotopic organization of the connectionist model does not
necessarily corresponds to the processing of visual inputs,
while it was initially inspired by the structure that has been
observed among the cortical columns of the visual areas.

Previous works of our team have shown that complex connectionist
models may be defined with a simple 2-dimensional architecture thanks
to configurable local connections (FPNA paradigm) [2,3]. The goal of these
post-doctoral researches is to propose methods to define
data-configured maps of retinotopically organized local neural
resources, so that a global functional behavior may emerge and
characterize the processed image. These researches will be carried
out with the help of the available works on FPNAs. Different aspects
will be addressed, such as the comparison of this approach with
emergent computations in bio-inspired models using
excitatory-inhibitory interactions [4], as well as the study
of the possible extensions of this approach in the field of
multimodal perception.

[1] C. Castellanos-Sanchez and B. Girau. Dynamic pursuit with a 
bio-inspired neural model.
In ACIVS 2005, volume 3708 of LNCS, pages 284--291, 2005.

[2] B. Girau. FPNA: Applications and implementations.
In A. Omondi and J. Rajapakse, editors, FPGA Implementations
 of Neural Networks. Kluwer, 2004.

[3] B. Girau. FPNA: Concepts and properties.
In A. Omondi and J. Rajapakse, editors, FPGA Implementations
 of Neural Networks. Kluwer, 2004.

[4] J. Vitay, N. Rougier and F. Alexandre. A distributed model of visual 
spatial attention.
Biomimetic Neural Learning for Intelligent Robotics, 2005.


Desired profile of candidat :

We aim at recruiting a post-doctoral fellow having a
strong background in connectionism, preferably applied to
visual perception. This work would also take advantage of
skills in software engineering (to integrate and widen
FPNA simulation tools) and in computational modeling.

Expected duration of fellowship:
12 months

Contact:
Bernard Girau
E-mail : girau at loria.fr
Tel : +33 3 83 59 20 58