突触权重修正及抑制性神经元数量对脑放电活动的影响

发布时间：2018-02-24 11:50

本文关键词： 神经网络突触可塑性 Izhikevich 神经元放电节律兴奋性神经元抑制性神经元　出处：《河北师范大学》2017年硕士论文　论文类型：学位论文

【摘要】：人脑含有上千亿的神经元细胞,有超过1410个神经突触,其结构与功能极为复杂,需要从不同层次对其进行探究。神经网络与全脑功能是神经信息学研究的重点,也是人类脑计划中的重要研究项目。突触是神经元之间信息传递的主要结构,它有在一定条件下增减数目、改变形态及调整功能的能力,即突触可塑性,大量研究证明突触可塑性是学习与记忆的细胞分子生物学基础。本文将通过神经网络模型从理论上研究突触可塑性修正参数、抑制性神经元所占比例对神经元放电活动的影响。本文研究内容分为以下三部分:第一部分研究了突触权重修正强度对神经网络放电活动的影响。可塑性神经网络中,单个神经元的动力学由Izhikevich模型决定,突触可塑性采用依赖神经元脉冲时序的可塑性规则(STDP),同时考虑了脉冲在轴突上的传导延迟。研究表明,此模型描述的放电活动中能出现?节律,?节律的呈现依赖突触权重的修正强度:对于突触权重增强修正的最大值(10)A,当??13.009.0(10)A时,网络在3600s的放电时间内存在?节律。其中当0.09 A0.1(10)??时,随着A(10)的增大,?节律首次出现的时间有所推迟,然而当0.1 A0.13(10)??时,随着(10)A的增大,?节律首次出现的时间有所提前;对于突触权重抑制修正的最大值-A,当??13.012.0-A时,网络在3600s的放电时间内存在?节律,并且随着-A的增大,?节律首次出现的时间有所提前。此部分内容在本文第二章。第二部分研究了描述权重变化快慢时间参数?对神经网络放电活动的影响。研究表明,当??217?时,网络在3600s的放电时间内存在?节律,并且随着?的减小,?节律首次出现的时间有所提前。此部分内容在本文第二章。第三部分研究了抑制性神经元所占比例对神经网络放电节律的影响。为了脑电平衡,神经网络中既有兴奋性神经元也存在一定数量的抑制性神经元。兴奋性神经元主要负责传递信息;抑制性神经元通过释放抑制性递质,使神经网络的活动始终处于一种稳定的状态。虽然抑制性神经元相比于兴奋性神经元的数目只是少量的,但它们对于神经网络中神经编码的演化是极其重要的。通过研究发现,?节律的产生依赖于抑制性神经元数量所占比例,当??174:826Ni:Ne200:800时(其中Ne:Ni为兴奋性神经元与抑制性神经元数量之比),网络在3600s的放电时间内才会存在?节律。此外给出了?节律首次出现的时间随抑制性神经元的数量所占比例的增减而发生的变化。在整个变化的过程中存在三个特殊的比例,通过对比这种情况下网络的放电神经元数目与时间的关系发现,大数量的放电事件比较稀疏时,它的放电频谱分布图中有?节律的产生,若大数量的放电事件比较密集时,它的放电频谱分布图中没有?节律的产生。
[Abstract]:The human brain contains hundreds of billions of neuronal cells, more than 1,410 synapses, the structure and function of which are extremely complex and need to be explored at different levels. Neural networks and whole brain functions are the focus of neuroinformatics. Synapse is the main structure of the transmission of information between neurons. It has the ability to increase and decrease the number, change shape and adjust function under certain conditions, that is, synaptic plasticity. A large number of studies have proved that synaptic plasticity is the cellular and molecular basis of learning and memory. The effect of the proportion of inhibitory neurons on the firing activity of neurons. This paper is divided into three parts: the first part studies the effect of the strength of synaptic weight correction on the discharge activity of neural networks. The dynamics of a single neuron is determined by the Izhikevich model. The synaptic plasticity is based on the plasticity rule of neuron pulse timing and the conduction delay of the pulse on the axon is taken into account. Rhythm? The presentation of rhythm depends on the correction intensity of synaptic weight: for the maximum value of synaptic weight enhancement correction, 10 A, when? ? At 13.009.0 ~ 10A, the network exists in the discharge time of 3600s? Rhythm. When 0. 09 A 0. 1 0. 0? ? Time, with the increase of A10)? The first time the rhythm appeared was delayed, however, when 0.1 A 0.13 + 10? ? With the increase of 10A? The first time of the rhythm appeared earlier; for the synaptic weight inhibition correction of the maximum A, when? ? At 13.012.0-A, the network exists in the discharge time of 3600s? Rhythm, and with the increase of A? The first time of the rhythm appears in advance. This part of the content in the second chapter. The second part of the study describes the weight changes of the fast and slow time parameters? Effects on the discharge activity of neural networks. ? 217? The network exists during the discharge time of 3600s. Rhythm, and with it? The reduction? In chapter 2, the effect of the proportion of inhibitory neurons on the discharge rhythm of neural networks was studied. There are not only excitatory neurons but also a certain number of inhibitory neurons in neural networks. Excitatory neurons are mainly responsible for transmitting information; inhibitory neurons release inhibitory transmitters. The activity of neural networks is always in a stable state. Although the number of inhibitory neurons compared to excitatory neurons is only a small number, they are extremely important for the evolution of neural coding in neural networks. The production of rhythms depends on the proportion of inhibitory neurons. ? 174: 826Ni2: 800 (where Ne:Ni is the ratio of the number of excitatory to inhibitory neurons, and the network does not exist until 3600s? Rhythm. Also given? The first time that the rhythm changes as the number of inhibitory neurons increases or decreases. There are three special proportions in the whole process. By comparing the relationship between the number of firing neurons and the time of the network, it is found that when a large number of discharge events are sparse, the discharge spectrum of the network has? When a large number of discharge events are dense, the rhythm is not found in the discharge spectrum. The production of rhythm.
【学位授予单位】：河北师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R338

【相似文献】