基于超結(jié)構(gòu)的BN隨機(jī)搜索學(xué)習(xí)算法
發(fā)布時(shí)間：2019-05-07 01:55
【摘要】：近年來,貝葉斯網(wǎng)絡(luò)(Bayesian network,BN)在不確定性知識(shí)表示與概率推理方面發(fā)揮著越來越重要的作用.其中,BN結(jié)構(gòu)學(xué)習(xí)是BN推理中的重要問題.然而,在當(dāng)前BN結(jié)構(gòu)的2階段混合學(xué)習(xí)算法中,大多存在一些問題:第1階段無向超結(jié)構(gòu)學(xué)習(xí)中存在容易丟失弱關(guān)系的邊的問題;第2階段的爬山搜索算法存在易陷入局部最優(yōu)的問題.針對(duì)這2個(gè)問題,首先采用Opt01ss算法學(xué)習(xí)超結(jié)構(gòu),盡可能地避免出現(xiàn)丟邊現(xiàn)象;然后給出基于超結(jié)構(gòu)的搜索算子,分析初始網(wǎng)絡(luò)的隨機(jī)選擇規(guī)則和對(duì)初始網(wǎng)絡(luò)隨機(jī)優(yōu)化策略,重點(diǎn)提出基于超結(jié)構(gòu)的隨機(jī)搜索的SSRandom結(jié)構(gòu)學(xué)習(xí)算法,該算法一定程度上可以很好地跳出局部最優(yōu)極值;最后在標(biāo)準(zhǔn)Survey,Asia,Sachs網(wǎng)絡(luò)上,通過靈敏性、特效性、歐幾里德距離和整體準(zhǔn)確率4個(gè)評(píng)價(jià)指標(biāo),并與已有3種混合學(xué)習(xí)算法的實(shí)驗(yàn)對(duì)比分析,驗(yàn)證了該學(xué)習(xí)算法的良好性能.
[Abstract]:In recent years, Bayesian networks (Bayesian network,BN) play an increasingly important role in uncertain knowledge representation and probabilistic reasoning. Among them, BN structure learning is an important problem in BN reasoning. However, in the current two-stage hybrid learning algorithms of BN structure, there are some problems: the first stage of undirected superstructure learning has the problem of easily losing the weak relations; The second stage of mountain climbing search algorithm is prone to fall into the local optimal problem. In order to solve these two problems, firstly, Opt01ss algorithm is used to study the superstructure, so as to avoid the phenomenon of losing edges as far as possible. Then the search operator based on superstructure is given, the random selection rule of the initial network and the stochastic optimization strategy of the initial network are analyzed, and the learning algorithm of SSRandom structure based on the random search of superstructure is put forward emphatically. To some extent, the algorithm can jump out of the local optimal extremum. Finally, on the standard Survey,Asia,Sachs network, through the sensitivity, specificity, Euclidean distance and the overall accuracy of four evaluation indicators, and compared with the existing three hybrid learning algorithm of the experimental analysis, to verify the good performance of the learning algorithm.
【作者單位】： 山西大學(xué)計(jì)算智能與中文信息處理教育部重點(diǎn)實(shí)驗(yàn)室;山西財(cái)經(jīng)大學(xué)信息管理學(xué)院;
【基金】：國家自然科學(xué)基金重點(diǎn)項(xiàng)目(61432011) 軍民共用重大研究計(jì)劃聯(lián)合基金重點(diǎn)項(xiàng)目(U1435212) 國家自然科學(xué)基金優(yōu)秀青年科學(xué)基金項(xiàng)目(61322211);國家自然科學(xué)基金項(xiàng)目(61672332) 中國博士后科學(xué)基金項(xiàng)目(2016M591409) 山西省自然科學(xué)基金項(xiàng)目(2013011016-4,2014011022-2)~~
【分類號(hào)】：TP18
本文編號(hào)：2470674
[Abstract]:In recent years, Bayesian networks (Bayesian network,BN) play an increasingly important role in uncertain knowledge representation and probabilistic reasoning. Among them, BN structure learning is an important problem in BN reasoning. However, in the current two-stage hybrid learning algorithms of BN structure, there are some problems: the first stage of undirected superstructure learning has the problem of easily losing the weak relations; The second stage of mountain climbing search algorithm is prone to fall into the local optimal problem. In order to solve these two problems, firstly, Opt01ss algorithm is used to study the superstructure, so as to avoid the phenomenon of losing edges as far as possible. Then the search operator based on superstructure is given, the random selection rule of the initial network and the stochastic optimization strategy of the initial network are analyzed, and the learning algorithm of SSRandom structure based on the random search of superstructure is put forward emphatically. To some extent, the algorithm can jump out of the local optimal extremum. Finally, on the standard Survey,Asia,Sachs network, through the sensitivity, specificity, Euclidean distance and the overall accuracy of four evaluation indicators, and compared with the existing three hybrid learning algorithm of the experimental analysis, to verify the good performance of the learning algorithm.
【作者單位】： 山西大學(xué)計(jì)算智能與中文信息處理教育部重點(diǎn)實(shí)驗(yàn)室;山西財(cái)經(jīng)大學(xué)信息管理學(xué)院;
【基金】：國家自然科學(xué)基金重點(diǎn)項(xiàng)目(61432011) 軍民共用重大研究計(jì)劃聯(lián)合基金重點(diǎn)項(xiàng)目(U1435212) 國家自然科學(xué)基金優(yōu)秀青年科學(xué)基金項(xiàng)目(61322211);國家自然科學(xué)基金項(xiàng)目(61672332) 中國博士后科學(xué)基金項(xiàng)目(2016M591409) 山西省自然科學(xué)基金項(xiàng)目(2013011016-4,2014011022-2)~~
【分類號(hào)】：TP18
【相似文獻(xiàn)】
相關(guān)期刊論文 前2條
1 王永皎;米浩坤;梁磊;;應(yīng)變和溫度變化對(duì)超結(jié)構(gòu)光纖光柵光譜的影響[J];光通信研究;2013年03期
2 李中科,楊曉輝,吳樂南;一種無需隨機(jī)搜索策略的自動(dòng)平面檢測方法(英文)[J];Journal of Southeast University(English Edition);2003年03期
相關(guān)碩士學(xué)位論文 前1條
1 武佳杰;基于超結(jié)構(gòu)與隨機(jī)搜索的BN結(jié)構(gòu)學(xué)習(xí)算法[D];山西財(cái)經(jīng)大學(xué);2017年
，本文編號(hào)：2470674
本文鏈接：http://sikaile.net/kejilunwen/zidonghuakongzhilunwen/2470674.html
最近更新
教材專著
