發(fā)布時(shí)間：2017-12-31 16:04

  本文關(guān)鍵詞：基于無(wú)監(jiān)督特征學(xué)習(xí)的演化計(jì)算行為分析　出處：《中國(guó)科學(xué)技術(shù)大學(xué)》2017年碩士論文　論文類型：學(xué)位論文

  更多相關(guān)文章： 演化計(jì)算 行為分析 無(wú)監(jiān)督學(xué)習(xí) 自組織映射 慢特征分析 深度信念網(wǎng)絡(luò) 受限玻爾茲曼機(jī)

【摘要】：演化計(jì)算作為一類啟發(fā)式優(yōu)化方法,其在解決真實(shí)世界中的復(fù)雜優(yōu)化問(wèn)題時(shí)的良好性能已經(jīng)在過(guò)去的幾十年中得到了很好的驗(yàn)證。但是演化計(jì)算自身復(fù)雜的隨機(jī)行為導(dǎo)致對(duì)其進(jìn)行理論分析異常困難,時(shí)至今日,仍然難以找到一種有效的方法來(lái)對(duì)演化算法在不同環(huán)境下的行為進(jìn)行學(xué)習(xí)和分析。為了更好地理解演化計(jì)算的行為,本文嘗試采用無(wú)監(jiān)督特征學(xué)習(xí)的方法,對(duì)演化計(jì)算在搜索過(guò)程中的一代群體行為進(jìn)行分析。首先對(duì)所研究的演化計(jì)算行為數(shù)據(jù)進(jìn)行定義,然后從基于自組織映射的演化計(jì)算行為數(shù)據(jù)預(yù)處理、基于慢特征分析的演化計(jì)算行為數(shù)據(jù)特征提取和基于深度信念網(wǎng)絡(luò)的演化計(jì)算行為數(shù)據(jù)特征提取三個(gè)方面入手,對(duì)演化計(jì)算的行為數(shù)據(jù)進(jìn)行了特征提取和分析。具體工作如下:1)研究了基于自組織映射的演化計(jì)算行為數(shù)據(jù)預(yù)處理方法。研究了基于t分布隨機(jī)鄰域嵌入(t-SNE)的自組織映射網(wǎng)絡(luò)預(yù)訓(xùn)練方法,從而將自組織映射網(wǎng)絡(luò)的訓(xùn)練分為二個(gè)階段:預(yù)訓(xùn)練、粗訓(xùn)練和微調(diào)三個(gè)階段,使得網(wǎng)絡(luò)能夠收斂到最好的狀態(tài)。然后使用訓(xùn)練好的自組織映射神經(jīng)網(wǎng)絡(luò)將原始高維空間中的演化計(jì)算行為數(shù)據(jù)映射到二維平面上,實(shí)現(xiàn)高維數(shù)據(jù)集的歸一化表示,為后續(xù)使用無(wú)監(jiān)督特征提取算法對(duì)演化計(jì)算行為數(shù)據(jù)進(jìn)行分析做好數(shù)據(jù)準(zhǔn)備。2)研究了基于慢特征分析算法的演化計(jì)算行為數(shù)據(jù)特征提取算法。首先對(duì)慢特征分析算法應(yīng)用到無(wú)監(jiān)督模式識(shí)別問(wèn)題時(shí)的時(shí)間序列結(jié)構(gòu)調(diào)整進(jìn)行了研究,同時(shí)對(duì)需要保留的慢特征維數(shù)也進(jìn)行了分析和計(jì)算。然后針對(duì)演化計(jì)算行為數(shù)據(jù)的特點(diǎn),設(shè)計(jì)了基于二階非線性擴(kuò)展慢特征分析算法的特征提取框架,對(duì)演化計(jì)算行為數(shù)據(jù)進(jìn)行特征提取。最后設(shè)計(jì)了多組對(duì)比實(shí)驗(yàn),分別研究了不同演化算法在同樣的landscape下的行為特征差異,以及同一演化算法在不同的landscape下的行為特征差異。實(shí)驗(yàn)結(jié)果表明,慢特征分析算法可以提取到不同演化算法之間具有判別性的穩(wěn)定特征。3)研究了基于深度信念網(wǎng)絡(luò)的演化計(jì)算行為特征提取算法。首先對(duì)深度信念網(wǎng)絡(luò)的基本組成單元——受限玻爾茲曼機(jī),進(jìn)行了詳細(xì)研究。然后針對(duì)要分析的演化計(jì)算行為數(shù)據(jù),設(shè)計(jì)了一個(gè)包含有七層受限玻爾茲曼機(jī)網(wǎng)絡(luò)的深度信念網(wǎng)絡(luò)框架。最后設(shè)計(jì)實(shí)驗(yàn)得到了不同演化算法在同一個(gè)測(cè)試函數(shù)下的行為數(shù)據(jù)經(jīng)過(guò)深度信念網(wǎng)絡(luò)提取到的特征分布結(jié)果,將該結(jié)果與慢特征分析提取到的特征進(jìn)行對(duì)比,對(duì)選用的四種演化算法的行為進(jìn)行了分析。
[Abstract]:Evolutionary computing is a kind of heuristic optimization method. Its good performance in solving complex optimization problems in the real world has been well verified in the past several ten years. However, the complex stochastic behavior of evolutionary computation makes it extremely difficult to theoretically analyze it. . Up to now, it is still difficult to find an effective way to study and analyze the behavior of evolutionary algorithms in different environments, in order to better understand the behavior of evolutionary computing. This paper attempts to use the unsupervised feature learning method to analyze the behavior of the generation of evolutionary computing in the search process. Firstly, the data of evolutionary computing behavior are defined. Then we preprocess the evolutionary behavior data based on self-organizing mapping. There are three aspects: feature extraction of evolutionary computing behavior data based on slow feature analysis and feature extraction of evolutionary computing behavior data based on deep belief network. The behavior data of evolutionary computing are extracted and analyzed. The main work is as follows: 1) the preprocessing method of evolutionary computing behavior data based on self-organizing mapping is studied, and the random neighborhood embedding based on t-distribution is studied. T-SNE-based self-organizing mapping network pretraining method. Thus, the self-organizing mapping network training is divided into two stages: pre-training, rough training and fine-tuning. The network can converge to the best state, and then use the trained self-organizing mapping neural network to map the evolutionary computing behavior data in the original high-dimensional space to the two-dimensional plane. The normalized representation of high-dimensional data sets is realized. Prepare the data for the analysis of evolutionary computing behavior data using unsupervised feature extraction algorithm. 2). The feature extraction algorithm of evolutionary computing behavior data based on slow feature analysis algorithm is studied. Firstly, the time series structure adjustment of slow feature analysis algorithm is studied when it is applied to unsupervised pattern recognition problem. At the same time, the dimension of slow feature which needs to be preserved is also analyzed and calculated. Then, a feature extraction framework based on second-order nonlinear extended slow feature analysis algorithm is designed according to the characteristics of evolutionary computing behavior data. Finally, we design a number of comparative experiments to study the behavior characteristics of different evolutionary algorithms under the same landscape. And the behavior characteristics of the same evolutionary algorithm under different landscape are different. The experimental results show that. Slow feature analysis algorithm can extract stable features with discriminant property between different evolutionary algorithms. 3). An evolutionary behavior feature extraction algorithm based on deep belief network is studied. Firstly, the constrained Boltzmann machine, which is the basic component of the deep belief network, is studied. Then the evolutionary behavior data to be analyzed are studied in detail. In this paper, we design a framework of deep belief network with seven layers of constrained Boltzmann machine network. Finally, we design experiments to obtain the behavior data of different evolutionary algorithms under the same test function, which are extracted by the deep belief network. Characteristic distribution results. The results are compared with the features extracted by slow feature analysis, and the behavior of the four evolutionary algorithms is analyzed.
【學(xué)位授予單位】：中國(guó)科學(xué)技術(shù)大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類號(hào)】：TP181

【參考文獻(xiàn)】

相關(guān)期刊論文 前1條

1 馬奎俊;韓彥軍;陶卿;王玨;;基于核的慢特征分析算法[J];模式識(shí)別與人工智能;2011年02期

相關(guān)博士學(xué)位論文 前1條

1 俞揚(yáng);演化計(jì)算理論分析與學(xué)習(xí)算法的研究[D];南京大學(xué);2011年

相關(guān)碩士學(xué)位論文 前1條

1 張雨晨;基于改進(jìn)的SVM和t-SNE高速列車走行部故障診斷[D];西南交通大學(xué);2016年

，

本文編號(hào)：1360459