Title :
Feed-forward neural network with adaptive buffer length
Author :
Tsunoda, Akihiro ; Hagiwara, Masafumi ; Nakagawa, Masao
Author_Institution :
Dept. of Electr. Eng., Keio Univ., Yokohama, Japan
Abstract :
Summary form only given, as follows. A feedforward neural network with adaptive buffer length (FAB) was proposed and simulated. FAB has buffers that store the past inputs, and switches between the buffers and hidden units to control the output of the buffers. Computer simulation results indicate that FAB can adjust the length of buffers for each input pattern automatically. FAB can also find the Markov dimension of the input sequences more effectively than the conventional dynamic network models
Keywords :
Markov processes; neural nets; self-adjusting systems; Markov dimension; adaptive buffer length; feedforward neural network; Adaptive systems; Automatic control; Buffer storage; Computational modeling; Feedforward neural networks; Feedforward systems; Multi-layer neural network; Neural networks; Shape; Switches;
Conference_Titel :
Neural Networks, 1991., IJCNN-91-Seattle International Joint Conference on
Conference_Location :
Seattle, WA
Print_ISBN :
0-7803-0164-1
DOI :
10.1109/IJCNN.1991.155624
