网状结构膜计算模型的计算理论及若干应用研究

Membrane computing is an emengent branch of natural computing, which aims to abstract novel computing ideas or models for solving computational hard problems from the structure and functioning of cells. Membrane computing models with net structure are distributed computing models which show a good ability of dealing with the information in parallel. One of the key problems for this model is to analysis its performance and use its good performance to solve the practical problems. In this project, on the basis of the work in the small exploration project supported by National Natural Science Foundation of China, we will investigate the impacts of some elelments of membrane computing models with net structure on its computational power, such as the woking mode, kind of cell or neuron, complexity of model. In this way, the theory of computational power of membrane computing models with net structure will be enriched. Such theory can be used for analysising the computational performance of membrane computing models with net structure in theory. As for the applications of membrane computing models with net structure, in this project we will learn from the results in applications obtained in neural network, DNA computing and membrane computing. Then, we will design several membrane optimization algorithms with net structure which own a better computational performance, and propose a new idea for the modeling and simulation of reticular biological systems, based on the structure and features of membrane computing models with net structure. In this project, we will not only introduce a novel high-performance computing model for the area of computer science, but also provide an innovative idea and method for the area of approximation calculation and the area of the modeling and simulation of biological systems.

膜计算是生物计算的新分支，旨在从细胞的结构和功能中获得解决计算复杂问题的计算思想或模型。网状结构膜计算模型是膜计算领域中一种具有良好并行处理信息能力的分布式计算模型。分析该模型的计算性能，并利用其良好计算性能解决实际问题是研究该模型的主要问题。本项目拟在国家自然科学基金小额探索项目的研究基础上，研究网状结构膜计算模型的工作模式、细胞或神经元类型、模型复杂度等因素对其计算能力的影响，完善其计算能力理论，为分析其计算性能提供理论依据。在应用方面，本项目拟借鉴神经网络、DNA计算、膜计算的应用成果，利用网状结构膜计算模型的结构及特点，建立具有良好计算性能的网状膜优化算法，提出网状生物系统建模与仿真的新思路。本项目的研究不仅为计算机科学引入一种新的高性能计算模型，同时也有望为近似计算和生物系统建模与仿真提供一种新的思路和方法。

生物系统是个复杂的计算系统，是获得解决计算复杂问题的思想、模型和算法的主要源泉。网状结构膜计算模型是该领域中一种具有良好并行处理信息能力的高性能计算模型，本项目针对此模型进行研究，主要研究内容包括网状结构膜计算模型的计算理论、优化算法的设计及生物系统建模与分析三个方面。本项目在此三方面取得了丰富的研究成果，发表了20多篇论文，其中多篇论文发表在《IEEE Transactions on Evolutionary Computation》、《IEEE Transactions on Neural Networks and Learning Systems》等期刊。在计算理论方面，本项目分别证明了网状结构膜计算模型的模型结构、神经元类型、所使用的规则类型与规则长度、神经元个数、模型工作模式等对其计算能力影响，完善了其计算能力理论。在优化算法设计方面，设计了基于knee point的高维多目标进化优化算法、多目标进化优化算法中的快速非支配排序算法、基于决策变量聚类的大规模高维多目标进化优化算法、高维多目标进化优化算法中的近似非支配排序算法、基于表层膜指导的多目标优化算法等多种高效算法，并开发了基于GPU 的优化算法实现平台。在生物系统建模与分析方面，建立了复杂网络的建模与基于膜结构的社团挖掘方法、基于随机游走的网络弱社团挖掘方法和基于网络分解的重叠社团挖掘方法。通过本项目的研究，为解决不同的计算复杂问题提供了多种高效的模型和算法。