Living Migration Algorithm And Its Application In Reconstructing The Gene Regulatory Network

Nature and human society, evolving endlessly, belong to a complex system. Inspired by the complicated bio-evolvement process and mechanism of nature, a good many Intelligent Optimization Algorithms (IOA) have been already proposed, such as Evolutionary Algorithms (EA), Simulated Annealing (SA), Tabu Search (TS), Ant Colony Optimization (ACO), Particle Swarm Optimization (PSO). These algorithms have been widely employed in many science researches and engineering fields.One of Darwin's researches—natural migration phenomena, especially about that human's activities of complex migration, conform another basic direction of the biology evolution. However, migration not only appears in the motion of macro-individuals, but also exists in the courses of microcosmic evolutions. In these courses, individuals or colonies of nature and human society migrate into other spaces or fit complex and unknown environment for searching the optimal living spaces. Attempting to abstract a kind of algorithms from the law of living and migrating, Living Migration Algorithm (LMA) is proposed, to enrich the function of IOA. And also given are the researches about LMA applied in the reconstruction of gene regulatory network for the sake of analysing the competence of LMA coping with the complex system.The main researches and innovations of the present thesis are as the following:1. A new adaptive and probabilitive optimization algorithm—LMA is firstly proposed based on the Darwin's bio-migration theory and the law of biologic living, as well as integrating simulating nature with imitating human society. The biologic background of LMA is given. Also expounded are the mechanism and the basic framework, as well as the implement course of LMA.2. Living Migration Computing (LMC) is given based on the thought of LMA, which is convenient for LMA employed better. The basic framework and the implement course of LMC are given too.3. The formalized and stochastic course description of LMA is given. It is proved that its operators belong to the selecting operator and the breeding one. Consequently, LMA is a kind of IOA. By means of the convergent definition, it is proved that LMA converges into the global optimization according to probability, and also shown are the convergent speed estimation and the complex analysis of LMA.4. The search performance of LMA is compared and analyzed in terms of carrying out LMA during four kinds of classical optimization problems. LMA is tested respectively during the convergent speed, the global optimization capability, the local subtle search ability, the search stability and the capacity of dealing with the noise. Experiment results indicate that LMA has much faster convergent speed, greater global optimization capability and better search stability. And LMA is able to deal with the optimization problem including noise.5. A new system structure of reconstructing gene regulatory network is proposed. LMA is applied on the reconstruction of Rat Central Nervous System (Rat CNS). And the analysis of the reconstructing results is given by means of the evaluation standard of gene regulatory network.