Human, Soheil, Bidabadi, Golnaz, Peschl, Markus. 2017. Learning to Satisfy Needs: Predictive Processing vs. Deep Learning. EUCognition, Zurich, Switzerland, 23.11.-24.11.

RIS

@MISC{Human2017,
title = {Learning to Satisfy Needs: Predictive Processing vs. Deep Learning},
author = {Soheil Human and Golnaz Bidabadi and Markus Peschl},
year = {2017},
address = {Zurich},
url = {https://www.eucognition.org/index.php?page=2017-zurich-programme},
language = {EN},
misc = {EUCognition},
abstract = {Need satisfaction has a key role in survival and wellbeing of biological cognitive agents. As a basic cognitive capability which presents early in development and throughout the lifespan, need satisfaction can be seen as an inspiring case for development of more human-like learning and thinking machines. In this paper, we first define the problem of need satisfaction and reformulate it as a learning problem. We then argue that deep learning is not an appropriate learning approach for development of computational cognitive models of need satisfaction. Finally, we introduce an alternative conceptual learning model for need satisfaction based on predictive processing (PP).},
}