Cognitive development in context: Learning to pay attention
Författare
Redaktör
- Jochen Triesch
- Tony Jebara
- Gedeon Deák
Summary, in English
A developing system must be able to learn new things
without forgetting what it has learned before. It should be
capable of reacting in different ways to the same stimuli in different contexts. Context sensitive reinforcement learning, which parallels some of the functions of the basal ganglia, is a learning algorithm that fulfills this requirement when the context is explicitly given. Here, we extend the algorithm with the ability to identify the relevant features of the environment that defines the context. It is suggested that this is a critical component of an architecture for cognitive development and we present simulation results that illustrate the operation of the system.
without forgetting what it has learned before. It should be
capable of reacting in different ways to the same stimuli in different contexts. Context sensitive reinforcement learning, which parallels some of the functions of the basal ganglia, is a learning algorithm that fulfills this requirement when the context is explicitly given. Here, we extend the algorithm with the ability to identify the relevant features of the environment that defines the context. It is suggested that this is a critical component of an architecture for cognitive development and we present simulation results that illustrate the operation of the system.
Avdelning/ar
Publiceringsår
2004
Språk
Engelska
Sidor
190-198
Publikation/Tidskrift/Serie
Proceedings of the International Conference on Development and Learning 04
Dokumenttyp
Konferensbidrag
Förlag
The Salk Institute for Biological Studies
Ämne
- Computer Vision and Robotics (Autonomous Systems)
- Learning
Status
Published
Projekt
- Ikaros: An infrastructure for system level modelling of the brain