Lattice Neural Networks for Incremental Learning
p. 107-120
Résumé
In incremental learning, it is necessary to conquer the dilemma of plasticity and stability. Because neural networks usually employ continuously distributed representation for state space, learning newly added data affects the existing memories. We apply a neural network with algebraic (lattice) structure to incremental learning, that has been proposed to model information processing in the dendrites of neurons. It has been proposed as a mathematical model of information processing in the dendrites of neurons. Because of the operation 'maximum' in lattice algebra weakening the continuously distributed representation, our proposed model succeeds in incremental learning.
Index
Texte
Citer cet article
Référence papier
Daisuke Uragami, Hiroyuki Ohta et Tatsuji Takahashi, « Lattice Neural Networks for Incremental Learning », CASYS, 24 | 2010, 107-120.
Référence électronique
Daisuke Uragami, Hiroyuki Ohta et Tatsuji Takahashi, « Lattice Neural Networks for Incremental Learning », CASYS [En ligne], 24 | 2010, mis en ligne le 06 September 2024, consulté le 20 September 2024. URL : http://popups.lib.uliege.be/1373-5411/index.php?id=3065
Auteurs
Daisuke Uragami
School of Computer Science, Tokyo University of Technology, 1404-1 Katakuramachi, Hachioji City, Tokyo 192-0982, Japan
Hiroyuki Ohta
Department of Physiology, National Defense Medical College, 3-2 Namiki, Tokorozawa, Saitama 359-8513, Japan
Tatsuji Takahashi
Division of Information System Design, School of Science and Technology, Tokyo, Denki University, Hatoyama, Hiki, Saitama 350-0394, Japan