Growing Neural Network

Summary
Neural networks are broadly used to approximate non-linear functions. However, it is difficult to decide an appropriate structure for a given task. Here, "Growing Neural Network" is proposed as an extention of Back Propagation (BP) learning. The propagated error signal is diffused from a target neuron as a substance. The axon of a growing neuron grows according to the concentration gradient of the substance and its activation. In a simulation, it was confirmed that the most simple problem such as "AND" and "OR" could be solved by the neural network and 2-layer structure was properly obtained.

Reference
3. Ryusuke Kurino, Masanori Sugisaka \& Katsunari Shibata:
Growing Neural Network for Acquisition of 2-layer Structure,
Proc. of IJCNN (Int'l Conf. on Neural Networks) 2003, 2003. 7 (to appear)

2. Ryusuke Kurino, Masanori Sugisaka \& Katsunari Shibata:
Acquisition of 2-layer Structure in a Growing Neural Network, Proc. of AROB (Int'l Symp. on Artificial Life and Robotics) 8th, pp. 391--394, 2003.1

1. 栗野竜輔, 杉坂政典, 柴田克成:
成長型ニューラルネットによる２層構造の獲得,
第21回計測自動制御学会九州支部学術講演会予稿集, pp. 245--248, 2002.12

Return to my home page (English)

Return to my home page (Japanese)