【摘要】：為深入研究生物大腦處理信息以及學(xué)習(xí)的能力,研究者們提出了人工神經(jīng)網(wǎng)絡(luò),用來(lái)模仿大腦信息表達(dá)以及處理的過程,而其中具有最高仿生性的是脈沖神經(jīng)網(wǎng)絡(luò),它表達(dá)信息以及處理信息均是采用對(duì)時(shí)間編碼的方式。比起感知機(jī)等傳統(tǒng)神經(jīng)網(wǎng)絡(luò),脈沖神經(jīng)網(wǎng)絡(luò)與生物大腦神經(jīng)元在信息處理機(jī)制方面更加接近。許多研究均表明,脈沖神經(jīng)網(wǎng)絡(luò)無(wú)論是在信息表達(dá)能力還是計(jì)算能力與傳統(tǒng)神經(jīng)網(wǎng)絡(luò)相比都更勝一籌。因而它引起了國(guó)內(nèi)外學(xué)者的廣泛關(guān)注和高度重視。目前,脈沖神經(jīng)網(wǎng)絡(luò)在人工智能等多方面領(lǐng)域已經(jīng)有一些初步研究成果,但是遠(yuǎn)沒達(dá)到商用的程度,相對(duì)傳統(tǒng)神經(jīng)網(wǎng)絡(luò)等它在實(shí)際應(yīng)用中還是較少的。首先是因?yàn)闀r(shí)間先后因素,研究相對(duì)并不是那么深入,也還沒有普及;再者,雖然脈沖神經(jīng)網(wǎng)絡(luò)被證實(shí)是與生物神經(jīng)系統(tǒng)最接近的網(wǎng)絡(luò),但是其生物大腦神經(jīng)系統(tǒng)的學(xué)習(xí)機(jī)制尚不清晰,對(duì)網(wǎng)絡(luò)中神經(jīng)元學(xué)習(xí)訓(xùn)練過程的研究也不成熟,因此,學(xué)習(xí)方法的研究目前仍然是一個(gè)值得研究的問題。為了充分運(yùn)用脈沖神經(jīng)網(wǎng)絡(luò)的優(yōu)點(diǎn),高度仿生性、較強(qiáng)的信息表達(dá)能力以及計(jì)算能力,本文對(duì)脈沖神經(jīng)網(wǎng)絡(luò)監(jiān)督學(xué)習(xí)算法進(jìn)行了深入的研究。目前已經(jīng)存在一些監(jiān)督學(xué)習(xí)算法,但是學(xué)習(xí)效率或者是適用性等方面還是不夠好,為了提高脈沖神經(jīng)網(wǎng)絡(luò)的學(xué)習(xí)效率、精確度以及能夠適應(yīng)更加復(fù)雜的問題,本文結(jié)合ReSuMe算法、SpikeProp算法等經(jīng)典算法或者規(guī)則,對(duì)多層網(wǎng)絡(luò)監(jiān)督學(xué)習(xí)算法進(jìn)行了優(yōu)化以及創(chuàng)新,并且對(duì)算法進(jìn)行了仿真和實(shí)驗(yàn)。本論文工作內(nèi)容如下:1)首先分析了目前存在的一些監(jiān)督學(xué)習(xí)算法性能、精確度等方面的優(yōu)缺點(diǎn),比如SpikeProp算法、ReSuMe算法等。2)提出一種多脈沖多層的神經(jīng)網(wǎng)絡(luò)監(jiān)督學(xué)習(xí)算法。該算法是結(jié)合ReSuMe算法,對(duì)目前存在的多層算法進(jìn)行優(yōu)化和創(chuàng)新,最后對(duì)算法進(jìn)行了仿真。3)在此基礎(chǔ)上,提出了基于延遲的神經(jīng)網(wǎng)絡(luò)監(jiān)督學(xué)習(xí)算法。該算法使得神經(jīng)網(wǎng)絡(luò)學(xué)習(xí)過程不再單單只是局限于突觸權(quán)重的調(diào)整,對(duì)原有的算法進(jìn)行了擴(kuò)展與創(chuàng)新,最后對(duì)算法進(jìn)行了仿真。4)最后將此算法成功應(yīng)用到XOR邏輯異或、Iris數(shù)據(jù)集分類等問題中,表現(xiàn)出了很好的效果。
[Abstract]:In order to study the ability of processing information and learning in the biological brain, researchers have proposed artificial neural networks, which are used to mimic the process of brain information expression and processing, and the most bionic is pulse neural network. It expresses information and processing information in a time coded way. The neural network, the pulse neural network and the biological brain neuron are closer to the information processing mechanism. Many studies have shown that the pulse neural network is better than the traditional neural network in both the information expression ability and the computing power. Therefore, it has aroused wide attention and attention of scholars at home and abroad. The pulse neural network has some preliminary research results in the field of artificial intelligence, but it is far from commercial. It is still less than the traditional neural network in practical application. First, because of the time successively, the research is relatively not so deep and has not been popularized; moreover, although the pulse neural network is not so popular. The network is proved to be the closest network to the biological nervous system, but the learning mechanism of the neural system of the biological brain is still not clear, and the study of the learning and training process of the neuron in the network is not mature. Therefore, the study of the learning method is still a problem worth studying at present. In this paper, there are some supervised learning algorithms, but the learning efficiency or applicability is not good enough, in order to improve the learning efficiency, accuracy and ability of the pulse neural network. In order to adapt to more complex problems, this paper combines ReSuMe algorithm, SpikeProp algorithm and other classical algorithms or rules to optimize and innovate the multi-layer network supervised learning algorithm, and the simulation and experiment of the algorithm are carried out. The work of this thesis is as follows: 1) first, the performance of some existing supervised learning algorithms is analyzed, and the accuracy of the algorithm is analyzed. The advantages and disadvantages of degree and other aspects, such as the SpikeProp algorithm, the ReSuMe algorithm and other.2), propose a multi pulse multilayer neural network supervised learning algorithm. This algorithm combines the ReSuMe algorithm to optimize and innovate the existing multi-layer algorithms. Finally, a simulation.3 is carried out on the algorithm. On the basis of this, a neural network monitoring system based on the delay is proposed. The algorithm makes the learning process of the neural network no longer only limited to the adjustment of the weight of the synapse, extends and innovating the original algorithm, and finally simulated the algorithm.4). Finally, the algorithm has been successfully applied to the XOR logic or Iris data collection class and so on, showing good results.
【學(xué)位授予單位】：電子科技大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類號(hào)】：TP18

【參考文獻(xiàn)】

相關(guān)期刊論文 前6條

1 曾曉勤;周建新;鄭星;鐘水明;;一種高效二進(jìn)前饋神經(jīng)網(wǎng)絡(luò)學(xué)習(xí)算法[J];哈爾濱工業(yè)大學(xué)學(xué)報(bào);2016年05期

2 方慧娟;羅繼亮;王飛;;基于學(xué)習(xí)速率自適應(yīng)的脈沖神經(jīng)網(wǎng)絡(luò)快速學(xué)習(xí)方(英文)[J];Chinese Journal of Chemical Engineering;2012年06期

3 王義萍;陳慶偉;胡維禮;;基底神經(jīng)節(jié)的尖峰神經(jīng)元網(wǎng)絡(luò)模型及其在機(jī)器人中的應(yīng)用[J];南京理工大學(xué)學(xué)報(bào)(自然科學(xué)版);2010年06期

4 蔡榮太;吳慶祥;;基于脈沖神經(jīng)網(wǎng)絡(luò)的邊緣檢測(cè)[J];微電子學(xué)與計(jì)算機(jī);2010年10期

5 張宏;方路平;童勤業(yè);;海豚等動(dòng)物神經(jīng)系統(tǒng)處理多普勒信號(hào)的一種可能性方案[J];物理學(xué)報(bào);2007年12期

6 彭建華;呂曉莉;劉延柱;;脈動(dòng)型神經(jīng)元網(wǎng)絡(luò)的聯(lián)想記憶與分割[J];計(jì)算力學(xué)學(xué)報(bào);2006年02期

相關(guān)碩士學(xué)位論文 前1條

1 鞏祖正;脈沖神經(jīng)網(wǎng)絡(luò)的多脈沖定時(shí)誤差反向傳播算法研究[D];西北師范大學(xué);2013年

，

本文編號(hào)：2157159