Artificial Development by Reinforcement Learning Can Benefit From Multiple Motivations

被引：6
作者：
Palm, Guenther [1 ]
Schwenker, Friedhelm [1 ]
机构：
[1] Ulm Univ, Inst Neural Informat Proc, Ulm, Germany
来源：
FRONTIERS IN ROBOTICS AND AI | 2019年 / 6卷
关键词：
reinforcement learning; multi-objective; actor-critic design; artificial curiosity; artificial cognition; intrinsic motivation; BOUNDED RATIONALITY; MODEL; FRAMEWORK; REWARDS; SYSTEMS; STATES; MAPS;
D O I：
10.3389/frobt.2019.00006
中图分类号：
TP24 [机器人技术];
学科分类号：
080202 ; 1405 ;
摘要：
Research on artificial development, reinforcement learning, and intrinsic motivations like curiosity could profit from the recently developed framework of multi-objective reinforcement learning. The combination of these ideas may lead to more realistic artificial models for life-long learning and goal directed behavior in animals and humans.
引用
收藏
页数：6
相关论文
共 95 条
[1]  
[Anonymous], FROM ANIMALS TO ANIMATS
[2]  
[Anonymous], COM ADAP SY
[3]  
[Anonymous], 2001, BAUPLAN SEELE
[4]  
[Anonymous], 2002, BOUNDED RATIONALITY
[5]  
Auer P., 2002, Journal of Machine Learning Research, V3, P397, DOI DOI 10.4271/610369
[6]  
Bach J., 2009, Principles of synthetic intelligence PSI: An architecture of motivated cognition
[7]  
Bach J., 2012, INT J SYNTHETIC EMOT, V3, P43, DOI [DOI 10.4018/JSE.2012010104, 10.4018/jse.2012010104]
[8]  
Barto A. G., 2004, INT C DEV LEARN EP R, P112
[9]  
Bellman R.E., 1957, DYNAMIC PROGRAMMING
[10]  
Bertsekas D. P., 1996, Neuro-Dynamic Programming
12345678910