Implementation And Research On Hopfield Neural Network In Associative Memory Storage

Hopfield neural networks also known as associative memory networks. Papers based on knowledge of Hopfield neural network and Eclipse plug-ins to construct a Hopfield neural network associative memory. It is able to associate a Clear message model integrity from Incomplete or vague information. Associative memory is the important functions of this memory, Can be used in Chinese, English, digital and other types of information in associative memory. Thesis used in Hopfield neural network knowledge, Eclipse plug-in mechanism for knowledge, XML-related knowledge. Eventually built that can be embedded into the eclipse development tools or other java plug-in system, a knowledge base can be expanded in the associative memory. According to incomplete information associated with the most similar complete information. Hopfield neural network described in the application of associative memory in the future and direction of development.Articles in the Hopfield neural network mechanisms and principles of in-depth study based on the Hopfield neural network related to the knowledge of the detailed description and comparison to master the core of the Hopfield neural network and design methods. In-depth study Hopfield neural network algorithm to solve TSP and other NP-complete problem, the proposed improved Hopfield neural network algorithm and its application to practice, the paper Hopfield neural network algorithm for the problems encountered in practical application, from reality, on the Hopfield neural network is improved. Gives an improved Hopfield neural network algorithm is applicable conditions of Hopfield neural network models of memory storage is greater than or much greater than the storage capacity of the perfect search To overcome the following limitations of Hopfield neural network algorithm:Hopfield neural network algorithm is not the ideal storage capacity problem, Hopfield neural network algorithm lack of inter-related, the pattern of memory and sometimes can not be too close to the Hopfield neural network algorithm recall a stored memory pattern mode is not any one model, but the drop into a "pseudo-state." And to improve the Hopfield neural network algorithm for the core design and practical the Hopfield associative memory. The memory can complete the fuzzy association information stored in the memory of complete and clear information.Improved Hopfield neural network algorithm has been improved in the following areas and enhance. First, Hopfield neural network algorithm for storage capacity is not ideal, namely, neural networks can store the maximum number of basic memory. Improved Hopfield neural network algorithm will split into smaller high-capacity storage capacity to solve the problem. Second, Hopfield neural network algorithm for the lack of inter-related, improved Hopfield neural network algorithm iterations into the system, according to stable if the noise model and the more similar the sample model number of the more iterations less able to think of a series of related memory. Again for the memory mode can not be too close. As the nature of associative memory is a class of optimization problems, which fall into non-existence of the normal local minimum possibility. Hopfield neural network algorithm improved by storing the pattern into blocks of memory, associative memory, avoiding the memory mode can not be too close to the problem. Finally, Hopfield neural network algorithm sometimes recall a stored memory pattern mode is not any one model, but rather fall into a pseudo-state. When the number of iterations is greater than a specified value, so that the system does not converge. Out of circulation, the next step associative memory stepsThe main work of the thesis:1) Recalling the history related to Hopfield neural networks and the development of Hopfield neural networks, Describes the theoretical innovation and the practical application of Hopfield neural networks. Hopfield neural network described in the status and impact of the field of artificial intelligence.2) Detailed analysis of the Hopfield neural network theory of knowledge. Including the continuous Hopfield network (Discrete Hopfield Network, DHN) of the definition and related knowledge, Hopfield network stability analysis and energy function, Hopfield network the number of neurons, neuron thresholds of neurons connecting strength values and so on.3) Eclipse plug-in mechanism for knowledge and XML for data storage knowledge to interact4) Design and Implementation of associative memory. Including the Hopfield neural network design and learning, Hopfield neural network algorithm are discussed in the practical application of problem, And propose an improved Hopfield neural network algorithm for associative memory.5) Introducing the actual association associative memory effects and the specific application. Including those for English letters associative memory, for digital or phone numbers associative memory, Chinese-language content associative memory and so on. Hopfield neural network to show associative memory and significance of the results of research, And will study in depth.