您的位置：首页 > 论文页面
基于回声状态神经网络的脑运动皮层神经信号的建模
发表时间：2009-09-30 浏览量：1736 下载量：572
| 全部作者： | 周汉英，王永骥 |
| 作者单位： | 华中科技大学控制科学与工程系；图像信息处理与智能控制教育部重点实验室 |
| 摘 要： | 回声状态神经网络（echo state networks, ESN）既是一种全新的回归神经网络，又是一种全新的学习方法。它的特点是通过大量神经元组成一个稀疏连接的网络，通过调整网络权值满足一定的条件从而使网络具有回声状态特性。它可以通过短期记忆能力存储历史的输入输出，达到很强的对非线性系统的动态逼近效应，而且学习算法简单。将其用于脑皮层神经信号的建模，输入信号是从猴子大脑运动皮层的神经元中取样的脉冲信号，而输出信号则是与之对应的猴子在实验过程中完成的朝虚拟空间8个方向运动的轨迹。实验结果显示：ESN能够很好地建立起猴子运动皮层信号与运动轨迹的实验模型。 |
| 关 键 词： | 计算机神经网络；建模；轨迹拟合；回声状态神经网络；脑机接口 |
| Title： | Modeling of brain motor cortical signals based on echo state networks |
| Author： | ZHOU Hanying, WANG Yongji |
| Organization： | Department of Control Science and Engineering, Huazhong University of Science and Technology; Key Laboratory for Image Processing & Intelligent Control, Education Ministry of China |
| Abstract： | Echo state network (ESN) is proposed as a novel recurrent neural network (RNN) model as well as a novel learning method. What distinguishes it from traditional RNNs is its dynamic reservoir which is comprised of a large number of processing elements which are sparsely connected with each other. Regulate the weights according to some simple constraints can guarantee the echo state property of the net which endues the net with the ability to save previous inputs and outputs, in this way this model can approximate nonlinear dynamic systems. Furthermore its training method is quite easy. In this paper, it is used for the modeling of brain cortical signals. The inputs are spiking signals recorded form the neurons in the motor cortex of a monkey while the outputs are the corresponding trajectories the monkey moves from the centre to one of eight targets in a 3D imaginary cube. The test results reveal that ESN can get satisfactory consequences in translating the monkey’s motor cortical signals into the desired trajectories. |
| Key words： | computational neural networks; modeling; trajectory fitting; echo state networks; brain machine interface |
| 发表期数： | 2009年9月第18期 |
| 引用格式： | 周汉英，王永骥. 基于回声状态神经网络的脑运动皮层神经信号的建模[J]. 中国科技论文在线精品论文，2009，2（18）：1937-1942. |
1
评论数 0请您登录
暂无评论
下载全文
多维论文