On Bootstrapping Sensori-Motor Patterns for a Constructivist Learning System in Continuous Environments

Abstract

The theory of cognitive development from Jean Piaget (1923) is a constructivist perspective of learning that has substantially influenced cognitive science domain. Indeed it seems that constructivism is a possible trail in order to overcome the limitations of classical techniques stemming from cognitivism or connectionism and create autonomous agents, fitted with strong adaptation ability within their environment, modelled on biological organisms. Potential applications concern intelligent agents in interaction with a complex environment, with objectives that cannot be predefined. There are numerous interesting works in developmental robotics going in this direction. In this work we investigate the application of these principles to a close domain: Ambient intelligence, which is extremely challenging but which also presents interesting aspects to exploit, like the participation of human users. From the perspective of a constructivist theory, the learning agent has to build a representation of the world that relies on the learning of sensori-motor patterns starting from its own experience only. This step is difficult to set up for systems evolving in continuous environments, using raw data from sensors without a priori modelling, primarily because they face a bootstrap problem. In this paper we address this particular issue and propose a decentralized approach based on a multi-agent framework, where the system's representations are constructed through a self-organization process that handles the dynamics between experience discretization and learning.

Efficient use of local energy

“…Finally, we would like to explore how this model can be used as a basis for artificial intelligence algorithms to manage real-time operations. All the information described is indeed very complementary to the type of information that a continuous management system such as those proposed by a company like Ubiant [16] would seek.…”

A Schema-Based Robot Controller Complying With the Constraints of Biological Systems

“…Computationally, it makes learning more difficult, unless Machine Learning is used for labeling this large space into a smaller set of relevant contexts. But this leads, in turn, to the bootstrapping problem (Drescher, 1991 ; Kuipers et al, 2006 ; Mazac et al, 2014 ), i.e., the fact that learning, if a context is relevant, requires to have labeled it in the first place. Through this example, we point out how the choice of the representation impacts the functional architecture of the system, by imposing specific mechanisms to deal with the problems arising from the chosen representation.…”

Smart Is a Matter of Context

“…Through message-parsing, the agents manage to trade electricity while satisfying the constraints of the grid in a decentralised way. Another example is the work of Mazac et al [18] which proposes to detect recurrent patterns at the interaction between a system and its environment. In this approach, inspired by the constructivist theory [21], three populations of agents interact with each other, guided by a feedback from the global system activity in order to construct relevant patterns and provide a model of the environment dynamic.…”