| Research Abstract |
In this research, we dealt with the attention phenomena of human perceptual-alternation and its quantum neuro computing model. In order to construct a new framework for describing the cohesiveness of the distribution and synthesis inherent in a neural network for this perceptual-alternation, a neural state is described quantum dynamically and a qubitlike neural network corresponding to the quantum circuit of quantum computation is studied. As a result, we construct the feed-forward neural network with our qubit neuron model, and discuss its information processing abilities with conventional neuron models through the learning simulations of the 4bit parity check problem and the general function identification problem, and we obtain the performance figure of learning. From these figures, we find that our qubit neuron model is more excellent than the conventional one. From our main results, we conclude that our qubit neuron model is an effective method of enhancing the performance of neural networks. We also construct a neural network model for perceptual alternation of ambiguous figures by using coupled oscillatory neurons or chaos neurons that can describe the timing of neuronal firing. From our results of simulations with this model, it is found that the model is able to represent the perceptual alternation in terms of the phase activities and synchronizations of neuronal phases. It is also shown that the distribution of perceptual duration fits Gamma distribution well as it is reported in psychophysical experiments. There, however, remains a further investigation of how to connect our qubit neuron with the coupled oscillatory perceptual-alternation model.
|
