New conditioning model for robots
Résumé
We present a neural network for the prediction of rewards in a conditioning model. It is based on two noisy-or and one noisy-and nodes and update rules inspired from BANNER technique. In specific cases, we show that the computation is similar to Rescorla-Wagner's equation, which inspired many computational models in the domain of conditioning
Domaines
Sciences cognitives
Origine : Fichiers produits par l'(les) auteur(s)
Loading...