Abstract
(English)
|
The BIBA project proposes a new model for perception, inference, learning and action called the Bayesian paradigm. This model is a completely new point of view on brain sciences, which could in the next years entirely change our conception of intelligence, computation and cognition. The project is organised along 3 axes of research and development:
- Neural basis of probabilistic inference.
- New probabilistic models and algorithms for perception and action.
- New probabilistic methodology and techniques, artefact conception ,and development. The expected results are new biological models and new methods and algorithms to design artefacts that adapt so as to act rationally with incomplete information.
OBJECTIVES
The twin scientific objectives of the BIBA project are:
- To reconsider in the light of Bayesian probabilistic reasoning our methodology, models, algorithms and techniques for building artefacts for the 'real world', gaining inspiration from the way living beings have evolved and adapted to the properties of their natural environments, and constructing artefacts that use these principles.
- To provide a firm Bayesian basis for understanding how biological systems use probabilistic logic to exploit the statistical properties of their environments, both at the level of neural mechanisms, and at the level of strategy, and to use artefacts to test the validity of these ideas.
DESCRIPTION OF WORK
The project is organised along 3 axes of research and development:
- Neural basis of probabilistic inference. We construct well-defined models of how probabilities may be represented and manipulated, and test predictions with psychophysical performance measures and studies of regional brain activation. We expect to improve our understanding of neural mechanisms and derive new ideas for the implementation of probabilistic inference in engineering systems.
- New probabilistic models and algorithms for perception and action. The main goal is to illustrate how probabilistic computation, and more specifically, Bayesian programming, may account for global behaviours of organisms in interaction with their environment. We will focus on specific questions concerning multi-sensory perception and motion control. We plan to develop new probabilistic models that explain the observed behaviours in human or animals and to implement them on autonomous artefacts.
- New probabilistic methodology and techniques or artefact conception and development. We will use the Bayesian paradigm to develop an artefact that acquires repertoires of reactive probabilistic behaviours (synergies), builds combinations, hierarchies and temporal sequences of these behaviours (strategies) and can be trained for different tasks. We shall imitate biology by reducing the amount of pre-specified knowledge of the environment built in by the designer, departing radically from classic robot design, and will explore the possibility for the artefact to 'discover' part of its preliminary knowledge using evolutionary techniques. We will evaluate the consequences of contraction theory. The result will be an artefact embodying efficient probabilistic algorithms for sensory interpretation and control, with more life-like behaviour and the means to test and develop further ideas.
|
