基于复杂系统的语言词频网络与脑情景记忆网络研究

发布时间：2018-01-31 13:13

本文关键词： 语言记忆复杂网络核磁共振成像脑神经网络管理神经学　出处：《上海大学》2016年博士论文　论文类型：学位论文

【摘要】：管理科学的发展离不开研究人的行为规律,人的行为又取决于大脑的运行过程。语言和记忆是人类对信息进行编码、储存和提取并再编码的外部表达和内部积累的两种认知功能。它们之间相互依存、协同运转。随着认知神经科学,脑成像技术,以及复杂系统理论的发展和成熟,人们日益重视总结脑、认知与管理之间的关系。本文基于复杂系统理论,以两条主线展开：构建并研究了语言文本中词汇不断生成的动态词频生长网络；通过先进的核磁共振成像技术,从人脑神经活动本质层面构建并分析了脑情景记忆网络。本文对复杂网络基本定律的理论研究,不仅可以应用在语言文本增长网络中,也可以应用于更大范围的人类活动量化模型中,可以从管理层面深层次理解大数据时代人类行为规律的统计特性,对网络演化进行分析和预测,具有经济、社会和学术意义。随着中国社会老龄化越来越严重,阿尔茨海默病作为威胁人类健康的四大疾病之一,具有不可治愈及发病隐匿性,发病人群逐渐年轻化。由于该病导致记忆和思维丧失,人格行为改变,无法进行正常决策,给社会和家庭造成了很大的精神、经济和时间负担。本文以该病为例,对该病早期阶段静息态情景记忆脑网络改变的研究具有理论、经济和社会意义。本文主要内容如下：(1)文章基于复杂系统理论对语言文本动态生成过程构建词频分布网络,对几种重要的复杂网络分布模型从理论层面进行了解释,厘清了它们的特性和相互之间关系。在西蒙模型择优选择机制的基础上讨论了语言学重要定律Heaps定律和复杂网络核心分布律幂律分布问的等价性关系。研究了一类幂律分布——-Waring分布与节点平均度随网络最大度增加呈次线性增长方式的充要条件。研究了幂律分布与致密性幂律(网络呈超线性增长)间的关系。给出了无标度函数的严格定义,证明了它的一个特性,对无标度网络进行了讨论。最后以长篇小说文本对上述理论模型进行了拟合与验证。(2)本文利用功能核磁共振成像技术,通过一种新的思路来构建静息态人脑情景记忆网络。验证了这种方法的合理性。通过研究该网络在阿尔茨海默病早期和后期阶段功能连接的变化,发现异常增加的功能连接可以预测记忆功能的衰退。通过小世界网络分析方法,对比了该病不同阶段网络小世界特性的变化,对发现的异常增加功能连接进行了辅助性解释。并测试了该小世界网络在随机故障和针对性攻击下网络的恢复能力。论文创新点在于：(1)从理论上用严格的数学过程证明了复杂系统几种重要定律间的相互关系,该结果可以扩展到更多地复杂网络实际应用中。(2)对使用上经常有分歧的“无标度网络”、“无标度函数”或“无标度分布”等概念从数学上进行了严格地定义与解释。(3)用真实实验研究方法对人格行为和认知功能改变从脑神经活动层面展开研究。用一种新的思路对静息态人脑情景记忆网络进行构建,克服了记忆功能在脑部不是一个独立功能,难以确定脑区的难点。用复杂网络分析方法研究了记忆网络结构的改变。(4)发现了静息态功能核磁共振成像在阿尔茨海默病早期情景记忆网络中功能连接的异常增加,这种增加预测了记忆功能的衰退,静息态成像具有操作简便应用范围广的特点。
[Abstract]:Behavior rules of the development of management science cannot do without the study of human, human behavior depends on the brain operation process. Language and memory is the information encoding human, two kinds of cognitive function and external expression of internal accumulation of storage and retrieval and encoding. They are interdependent, cooperative operation. With cognitive neuroscience that brain imaging techniques, and the development of complex system theory and mature, people pay more attention to summarize the relationship between brain, cognition and management. This paper based on the theory of complex systems in two aspects: to construct and study the dynamic frequency of vocabulary has been generating language text in the growth of network; by magnetic resonance imaging technology, construction from the human nature of neural activity and analysis of brain episodic memory network. In this paper, theoretical research on the basic law of complex network, not only can be used in the growth of net language text In a network, a quantitative model of human activities can also be applied to a wider range of statistics can understand big data era of human behavior from the aspect of management deeply, with economic analysis and prediction of network evolution, and social and academic significance. With the aging society China of more and more serious, as one of the four Alzheimer's disease the big threat to human health, which can not be cured and the incidence of occult onset, the crowd gradually younger. Because the disease leads to loss of memory and thinking, personality and behavior change, unable to carry out normal decision-making, caused a great spirit to the society and family, economic burden and time. This paper takes the disease as an example, with theory in the early stages of the resting state brain episodic memory network changes of the disease, economic and social significance. The main contents of this paper are as follows: (1) based on the theory of complex system of dynamic language text Construction of frequency distribution network state generation process, several important complex network distribution model is explained from the aspect of theory, clarifies the relation between the characteristics of them and each other. Based on the Simon model selection mechanism on linguistics important law Heaps law and the complex network core distribution of power-law distribution of equivalence relation Q. A power-law distribution, -Waring distribution and the average degree of nodes with the maximum degree increasing network linear growth mode of the necessary and sufficient conditions. Study the power-law distribution and density of the power-law (network is superlinear). The relationship between the given strict definition of scaling function, prove a the characteristics of it, the scale-free network is discussed. Finally, the novel text of the theoretical model are fitted and verified. (2) using functional magnetic resonance imaging, through a A new idea to construct the resting state of episodic memory network. To verify the rationality of this method. By studying the changes of the network connection in Alzheimer's disease early and late stage function, found abnormal recession increased functional connectivity can predict memory function. Through the small world network analysis method, comparison of changes in different stages of the disease network the characteristics of small world, to find the abnormal increase of functional connectivity of auxiliary interpretation. And testing of this small world network with random and targeted attacks under the network fault recovery capability. Innovation of this paper lies in: (1) proved the relationship between several important law of complex system theoretically using mathematical process strictly the results can be extended to the application of more complex networks. (2) to often have differences using the "scale-free network", "scaling function" or "no The scale distribution "and other concepts were strictly defined and explained mathematically. (3) studied from the level of brain activity on personality behavior and cognitive function change with the actual experimental research method. In the resting state of episodic memory network construction in a new way, to overcome the memory function of the brain is not a a separate function, it is difficult to determine the difficulty of brain regions. Study memory network structure change by using complex network analysis method. (4) found a resting state functional magnetic resonance imaging in Alzheimer's disease early functional episodic memory network connection in the abnormal increase, this increase predicted its memory function, resting state imaging has the characteristics of operation is a wide range of applications.

【学位授予单位】：上海大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：H030

【相似文献】