Learning and Selecting Behaviours with Dynamic Neural Fields

old_uid15972
titleLearning and Selecting Behaviours with Dynamic Neural Fields
start_date2015/07/21
schedule10h
onlineno
location_info5e étage, salle 570
summaryDynamic Neural Fields are neurally-inspired dynamical systems, which allow to model cognitive functions, such as memory formation, decision making, stabilisation of control parameters, feature binding, and executive control. In this talk, I will describe how DNFs were extended recently into a structural representation of intentional behaviours, which features a representation of the intention and the condition of satisfaction of each action. I will demonstrate how this new structure allows to learn behaviors autonomously and to organise their activation and deactivation in an appropriate sequence, leading to a behavioural goal.
responsiblesMarinica