Title :
Study on effect of deterministic learning sequencer
Author :
Lursinsap, Chidchanok ; Watanapongse, Pirawat
Author_Institution :
Dept. of Math., Chulalongkorn Univ., Bangkok, Thailand
fDate :
27 Jun-2 Jul 1994
Abstract :
Our hypothesis in the paper is that the learning speed of our brain depends heavily on how the sequence of information is consumed. We show that the deterministic feeding sequence of training vectors plays a very significant role in speeding up the learning process. This sequence controls the learning rate and the initial weight vectors as well as convergence speed. We focus the hypothesis on only the feedforward networks in binary space. The mathematical relation between weight vectors, learning rate, and the training vector sequence are derived. The experiments are conducted with training vectors in binary field
Keywords :
feedforward neural nets; learning (artificial intelligence); binary field; binary space; brain; convergence speed; deterministic feeding sequence; deterministic learning sequencer; feedforward networks; initial weight vectors; learning rate; learning speed; mathematical relation; training vector sequence; weight vectors; Biological neural networks; Convergence; Error correction; Feedforward neural networks; Least squares methods; Mathematics; Neural networks; Neurons; Process control; USA Councils;
Conference_Titel :
Neural Networks, 1994. IEEE World Congress on Computational Intelligence., 1994 IEEE International Conference on
Conference_Location :
Orlando, FL
Print_ISBN :
0-7803-1901-X
DOI :
10.1109/ICNN.1994.374175