The DN-SARSA(λ) model provides a framework which shows how computational learning algorithms can be incorporated into a continuous neural-dynamical model. This enables autonomous learning and acting in continuous and dynamic environments, a challenge that is easily overlooked when formalizing the learning problem in discretized spaces without accounting for their coupling to sensory-motor dynamics. © 2012 IEEE.
|Original language||English (US)|
|Title of host publication||2012 IEEE International Conference on Development and Learning and Epigenetic Robotics, ICDL 2012|
|State||Published - Dec 1 2012|