基于生物启发计算的知识发现关键技术研究与实现-计算机应用技术专业论文.docx

四川大学博士学位论文警模型． 四川大学博士学位论文 警模型． 本文在基于生物启发计算的知识发现关键技术研究中提出并实现了一些新 的算法，并对算法作了理论和实验分析，与传统方法相比，在速度、精度和成 功率上都有明显提高。同时，本文将这些算法引入到一些具体应用中，发挥了 实际的成效。本文的工作对于研制新型的智能系统具有重要的理论意义和应用 前景。 关键词：生物启发计算智能计算遗传算法神经网络概念相似度基因表 达式编程 —田卜 基于生物启发计算的知识发现关键技术研究与实现Bio-inspired 基于生物启发计算的知识发现关键技术研究与实现 Bio-inspired Computing Based Key Technology Researches and Implementations of Knowledge Discovery Computer Application Technology Ph．D Candidate PENG Jing Advisor TANG Chang-jie Bio·-inspired Computing is the intelligence··computing model enlightened by natural intelligence and biological processes．In recent years，Bio-inspired Computing has been applied widely in many fields，such as artificial intelligence，machine learning and data mining．Enlightened from new achieves in biological science,computer scientists have proposed many new theories and algorithms of bio·inspired computing while it is still in the beginning stage comparing with other elder science fields． This dissertation has explored some interesting and challenging issues in bio-inspired computing．The main contributions include： 1．Proposes a new evolution algorithm：M—GEP(Multi—Layer Chromosome Gene Expression Programming)based on the new concept named multi-layer chromosomes．The algorithm is efficient in real applications，such as function discovery,electronic circuit evolution,ete．Extensive experiments on the traditional single gene and multi—genes GEP show that the average number of generations of M．GEP iS reduced to 29％～8l％． 2．Proposes and implements a new evolution algorithm-MEOE(multi—gene evolutionary algorithm based on overlapped expression)．Biologists find that nucleoside sequence,which forms the genes．Can be overlapped as overlapping gene in specific conditions．Biologists believe that overlapping gene can．save material and energy to store genetic information；moreover,it Can adjust the expression of genes．Enlightened by overlapping gene and concept ofthickness in natural inlmune system,the dissertation proposes and implements a new