Institute of Cognitive Science, University of Osnabrück, Wachsbleiche 27, 49090 Osnabrück, Germany.
Institute of Cognitive Science, University of Osnabrück, Wachsbleiche 27, 49090 Osnabrück, Germany; Institute of Neurophysiology and Pathophysiology, University Medical Center Hamburg-Eppendorf, Hamburg, Germany.
Neural Netw. 2021 Feb;134:23-41. doi: 10.1016/j.neunet.2020.11.004. Epub 2020 Nov 23.
How do humans acquire a meaningful understanding of the world with little to no supervision or semantic labels provided by the environment? Here we investigate embodiment with a closed loop between action and perception as one key component in this process. We take a close look at the representations learned by a deep reinforcement learning agent that is trained with high-dimensional visual observations collected in a 3D environment with very sparse rewards. We show that this agent learns stable representations of meaningful concepts such as doors without receiving any semantic labels. Our results show that the agent learns to represent the action relevant information, extracted from a simulated camera stream, in a wide variety of sparse activation patterns. The quality of the representations learned shows the strength of embodied learning and its advantages over fully supervised approaches.
人类如何在几乎没有环境提供的监督或语义标签的情况下，获得对世界的有意义的理解？在这里，我们研究了行动和感知之间的闭环作为这个过程中的一个关键组成部分的体现。我们仔细研究了一个深度强化学习代理所学到的表示，该代理是使用在具有非常稀疏奖励的 3D 环境中收集的高维视觉观察进行训练的。我们表明，该代理在没有收到任何语义标签的情况下，学会了对有意义的概念（如门）的稳定表示。我们的结果表明，代理学会了以各种稀疏激活模式来表示从模拟相机流中提取的与动作相关的信息。所学到的表示的质量显示了体现学习的优势，以及它相对于完全监督方法的优势。
