论文部分内容阅读
针对智能体的行为认知问题,提出一种小脑与基底神经节相互协调的行为认知计算模型.该模型核心为操作条件学习算法,包括评价机制、行为选择机制、取向机制及小脑与基底神经节的协调机制.初期的学习信号来自于下橄榄体和黑质两部分,在熵的意义上说明该算法是收敛的.采用该学习方法为自平衡两轮机器人建立运动神经认知系统,利用RBF网络逼近行为和评价网络.仿真实验表明该方法改善仅有基底神经节作用的行为-评价算法学习速度慢和失败次数多的问题,学习后期通过温度的不断降低,加快学习速度,震荡逐渐消失,改善学习效果.
Aiming at the problem of behavior cognition of agent, this paper proposes a behavioral cognition calculation model of coordination between cerebellum and basal ganglia.The core of the model is the learning algorithm of operating conditions, including evaluation mechanism, behavior selection mechanism, orientation mechanism and cerebellum and basal ganglia Section coordination mechanism.The initial learning signal comes from the lower olivin and the substantia nigra.It indicates that the algorithm is convergent in the sense of entropy.Using this learning method to establish a motor nerve cognitive system for self-balancing two-wheeled robots, RBF network approximation behavior and evaluation network.The simulation experiments show that this method can improve the behavior of only the basal ganglia - the evaluation algorithm slow learning and the failure of the number of more problems, learning through the continuous reduction of temperature to speed up the learning speed, oscillation disappear , Improve learning outcomes.