Abstract
This paper presents an N2 neuron N-parallel computation method for a neural network for the N queens problem. In this method, the N2 neurons are partitioned into N groups of N neurons, where the states of N neurons in N different groups are updated synchronously and the states in the same group are updated sequentially. First, we compare the performance of two existing neural networks for the N queens problem in order to show that Takefuji's neural network is best in the sequential method. Then, we show that even the improved neural network using the N2-parallel method cannot provide the same performance as that using the using sequential method. Lastly, we propose an N-parallel method to achieve high performance and parallel computation simultaneously.
Original language | English |
---|---|
Pages (from-to) | 12-20 |
Number of pages | 9 |
Journal | Electronics and Communications in Japan, Part III: Fundamental Electronic Science (English translation of Denshi Tsushin Gakkai Ronbunshi) |
Volume | 80 |
Issue number | 11 |
Publication status | Published - Nov 1997 |
Externally published | Yes |
Keywords
- N-parallel computation
- N-queens problem
- Neural network
ASJC Scopus subject areas
- Electrical and Electronic Engineering