电讯技术
電訊技術
전신기술
TELECOMMUNICATIONS ENGINEERING
2013年
4期
402-407
,共6页
非线性独立分量分析%扩展互信息分离算法%多层感知机%双隐层神经网络
非線性獨立分量分析%擴展互信息分離算法%多層感知機%雙隱層神經網絡
비선성독립분량분석%확전호신식분리산법%다층감지궤%쌍은층신경망락
nonlinear independent component analysis%extended mutual information separation algorithm%multi-layer perception%double hidden layer neural network
扩展互信息分离算法采用单隐层神经网络近似算法代价函数中的非线性函数,可调节的参数有限,需要多次迭代才能收敛,从而导致收敛速度较慢.针对这一问题,采用双隐层神经网络近似非线性函数,以分离结果的互信息最小化作为代价函数,采用梯度下降方法对代价函数进行优化,增加了可调节参数数量.仿真实验结果表明,改进后的算法相对原算法收敛速度更快,误差更小.
擴展互信息分離算法採用單隱層神經網絡近似算法代價函數中的非線性函數,可調節的參數有限,需要多次迭代纔能收斂,從而導緻收斂速度較慢.針對這一問題,採用雙隱層神經網絡近似非線性函數,以分離結果的互信息最小化作為代價函數,採用梯度下降方法對代價函數進行優化,增加瞭可調節參數數量.倣真實驗結果錶明,改進後的算法相對原算法收斂速度更快,誤差更小.
확전호신식분리산법채용단은층신경망락근사산법대개함수중적비선성함수,가조절적삼수유한,수요다차질대재능수렴,종이도치수렴속도교만.침대저일문제,채용쌍은층신경망락근사비선성함수,이분리결과적호신식최소화작위대개함수,채용제도하강방법대대개함수진행우화,증가료가조절삼수수량.방진실험결과표명,개진후적산법상대원산법수렴속도경쾌,오차경소.
Extended mutual information separation(EMISEP)algorithm uses a single hidden layer neural network to approximate nonlinear function of cost function,so the adjustable parameter is limited and it needs more itera-tion times to converge,which leads to relatively slow convergence speed. To overcome this problem,this paper uses double hidden layer perceptions to approximate nonlinear function of cost function,and uses mutual infor-mation minimum of separation signals as cost function,which is optimized by gradient descent method. This in-creases the number of adjustable parameters. The simulation results prove that the improved algorithm has faster convergence speed and smaller error comparing with the original algorithm.
