基于正交校正共軛梯度法的快速神經(jīng)網(wǎng)絡(luò)學(xué)習(xí)算法研究
Study of fast learning algorithm for neural networks base on CGM-OC approach
-
摘要: 前饋神經(jīng)網(wǎng)絡(luò)由于具有理論上逼近任意非線性連續(xù)映射的能力,因而非常適合于非線性系統(tǒng)建模及構(gòu)成自適應(yīng)控制。為了提高前饋神經(jīng)網(wǎng)絡(luò)的權(quán)的學(xué)習(xí)效率及穩(wěn)定性,該文提出一種基于正交校正共軛梯度優(yōu)化方法的快速神經(jīng)網(wǎng)絡(luò)學(xué)習(xí)算法,通過與其它學(xué)習(xí)算法(如:BP算法、變尺度法、用差商近似代替導(dǎo)數(shù)的Powell法等)的比較,經(jīng)仿真試驗(yàn)表明,本算法是一種高效、快速的學(xué)習(xí)算法。Abstract: Because the feedforward neural network has an ability of approach to arbitrary nonlinear mapping, it can be used effectively in the modeling and controlling of nonlinear system. In order to improve the learning efficiency and stability of feedforward neural network, a fast learning algorithm for neural networks base on CGM-OC approach is presented. Compared with other learning methods such as BP method, Broyden Flecher Goldfarl Shanno method. Power method etc., simulation results show that the proposed method is an efficient and fast method.
-
徐嗣鑫,戴友元,前向神經(jīng)網(wǎng)絡(luò)的一種快速學(xué)習(xí)算法及其應(yīng)用,控制與決策,1993,8(4),284-288[2]王耀南,童調(diào)生,蔡自興,基于神經(jīng)元網(wǎng)絡(luò)的智能PID控制及應(yīng)用,信息與控制,1994,23(3),185-189[3]張星昌,前饋神經(jīng)網(wǎng)絡(luò)的新學(xué)習(xí)算法研究及其應(yīng)用,控制與決策,1997,12(3),213-216[4]王耀南,基于變尺度優(yōu)化方法的快速神經(jīng)網(wǎng)絡(luò)學(xué)習(xí)算法研究,系統(tǒng)仿真學(xué)報(bào),1997,9(1),34-39.[5]徐春暉,徐向東,前饋神經(jīng)網(wǎng)絡(luò)新學(xué)習(xí)算法的研究,清華大學(xué)學(xué)報(bào)(自然科學(xué)版)1999,39(3),1-3. -
計(jì)量
- 文章訪問數(shù): 2474
- HTML全文瀏覽量: 117
- PDF下載量: 445
- 被引次數(shù): 0
下載:
