Research On Associative Memory Properties Of The Chaotic Neural Network

Increasing of information demands computer having intelligent ability in the field of message treatment, in which exploring a novel method of associative memory is a very important topic. The chaotic neural network which combines the features of chaos such as randomicity, going through all over the states and sensibility to initial value with the features of neural network such as large scale parallel of processing, distributing type of message storage, associative memory and robustness is a novel method of associative memory. Based on other research, the author pays his main attention to the feature of associative memory for chaotic neural network and researches'on the following three aspects: I ) Research on the parameter effecting on associative memory properties. II) Research on the structure effecting on many-to-many associative memory properties. III)Research on the successive learning ability.The paper is divided into three parts. Firstly, from chapter one to chapter two the author mainly states the importance of combining the chaos with the neural network in the field of the intelligent message treatment, especially of associative memory. Moreover, the author briefly discusses the theoretical basement of the chaotic neural network. Secondly, from chapter three to chapter five the author researches on the effect of parameter and structure on the chaotic neural networks and its successive learning capability. Through study the author obtainssome valuable efforts. Lastly, the author summarizes the paper and prospects to the further research fields.In short, through researches the author discovers chaotic neural networks having tremendous preponderance in the field of intelligent message treatment, particularly of separation of superimposed pattern, many-to-many associations and successive learning.